<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0"
    xmlns:content="http://purl.org/rss/1.0/modules/content/"
    xmlns:dc="http://purl.org/dc/elements/1.1/"
    xmlns:atom="http://www.w3.org/2005/Atom">
    <channel>
        <title>wrike-ai — Wrike Community</title>
        <link>https://community.wrike.com/</link>
        <pubDate>Mon, 06 Apr 2026 06:25:16 +0000</pubDate>
        <language>en</language>
            <description>wrike-ai — Wrike Community</description>
    <atom:link href="https://community.wrike.com/discussions/tagged/wrike-ai/feed.rss" rel="self" type="application/rss+xml"/>
    <item>
        <title>✔️ AI Agents March Update: Precision Control — Route, Filter, and Track</title>
        <link>https://community.wrike.com/discussion/2829/%EF%B8%8F-ai-agents-march-update-precision-control-route-filter-and-track</link>
        <pubDate>Fri, 20 Mar 2026 05:35:12 +0000</pubDate>
        <category>Wrike AI</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2829@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This release is all about precision and control. Agents can now move or add tasks to specific folders, adjust start and due dates, and reason across context (parent fields, sibling tasks, cross-folder lookups). They trigger on API changes, support per-action filters and checkbox fields, and can read formula field outputs. Featured example: an “intake deduplication” agent that scans existing work for similar items, flags likely duplicates, and routes only truly new requests, replacing complex rule sets with a single reasoning agent.</p><p><br />
Hey Community! 👋</p><p>Welcome to <strong>Wrike Agent News</strong>!</p><p>This monthly digest features updates, learnings, and how-tos from the Wrike R&amp;D team.</p><p>This month’s theme is <strong>precision</strong>: every new feature is focused on giving you finer control over what agents do, where they operate, and how you track everything that happens.</p><h2 data-id="new-action-location-change"><strong>📍 New Action: Location Change </strong></h2><p>Agents can now move tasks between folders or add them to additional folders based on content, status, or whatever logic you put in the prompt.</p><p><strong>There are two modes:</strong></p><ul><li><strong>Move to location</strong> pulls the task out of its current folder and places it exclusively in the target.</li><li><strong>Add to location</strong> keeps it where it is and adds it to another folder, which is useful when you want visibility across teams without losing the original context.</li></ul><p>When writing the agent’s instructions, you embed location chips (structured references to specific folders or projects) directly in the prompt text. The agent reads the task, reasons about which destination fits, and picks from those chips. You control where things can land; the agent decides which one.</p><p><strong>Why this matters:</strong> Imagine a setup with more than 15 automation rules just to route tasks to the right folder, each rule covering a single condition. With location change, a single agent can read the task, reason about where it belongs, and move it. One agent replaces the whole rule set because the LLM handles the combinatorial logic that static if‑then rules cannot.</p><h2 data-id="new-action-start-and-due-date-change"><strong>📆 New Action: Start and Due Date Change </strong></h2><p>Agents can now modify start dates and due dates based on workflow conditions.</p><p>Use case: A project kickoff agent can read the project start date, calculate milestone dates based on the project type, and set due dates on key deliverables. Or a simpler pattern: when a task moves to “In Progress,” set the start date to today if it’s empty.</p><p><strong>Why this matters:</strong> Date management has been one of the most requested agent capabilities, and it is an area where agents add clear value over basic automations. An automation can set a fixed date, but an agent can read the task context and reason about what the right date should be.</p><h2 data-id="expanded-context-parent-fields-siblings-and-cross-folder-lookup"><strong>🔎 Expanded Context: Parent Fields, Siblings, and Cross-Folder Lookup </strong></h2><p>Agents just got significantly smarter about what they can see.</p><h4 data-id="parent-item-custom-fields"><strong>Parent item custom fields</strong></h4><p>Agents acting on a sub-task can now read custom field values from the parent item. Use case: a task inherits its priority from the project it belongs to, so the agent reads the project’s “Client Tier” field and uses it to decide how to handle the sub-task.</p><h4 data-id="sibling-awareness"><strong>Sibling awareness</strong></h4><p>Agents can now see other sub-tasks under the same parent. This enables duplicate detection (is there already a task like this one?), workload balancing (how many open tasks does this folder have?), and comparative reasoning (is this the only overdue item, or are all of them late?).</p><h4 data-id="cross-folder-lookup"><strong>Cross-folder lookup</strong></h4><p>Agents can read items from other folders in the same space, not just the folder they are appointed to. Example: a new project request comes into an intake folder, and the agent scans a “Forecasted Projects” folder to check if a matching project already exists before creating a duplicate.</p><p>Together, these upgrades let agents reason about context beyond the individual task they were triggered on, shifting from “react to one item” to “understand the situation.”</p><h2 data-id="agents-now-fire-on-api-changes"><strong>🔗 Agents Now Fire on API Changes </strong></h2><p>Agents now trigger when items are created or modified via the Wrike API. Previously, only UI-driven changes fired triggers.</p><p><strong>Why this matters:</strong> If you use Wrike Integrate, Zapier, or any custom integration that creates or updates tasks via API, your agents will now pick those changes up automatically.</p><h2 data-id="easier-agent-discovery-comment-icon"><strong>🏷️ Easier Agent Discovery: Comment Icon </strong></h2><p>There’s a new icon in the comment editor toolbar for finding and appointing agents. Instead of navigating to the agent management panel, you can discover available agents directly from the comment area of any work item.</p><p>This also improves the chat experience: selecting an already appointed agent from the comment bar starts a conversation with it right there in the thread.</p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/XHP70UIPQJZT\/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png&quot;,&quot;name&quot;:&quot;Screenshot 2026-03-18 at 11.02.19 AM.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:11087,&quot;width&quot;:440,&quot;height&quot;:103,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FXHP70UIPQJZT%2Fscreenshot-2026-03-18-at-11-02-19-e2-80-afam.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3650,&quot;dateInserted&quot;:&quot;2026-03-20T05:32:27+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FXHP70UIPQJZT%2Fscreenshot-2026-03-18-at-11-02-19-e2-80-afam.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png" alt="Screenshot 2026-03-18 at 11.02.19 AM.png" height="103" width="440" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png 2000w, https://us.v-cdn.net/6038595/uploads/XHP70UIPQJZT/screenshot-2026-03-18-at-11-02-19-e2-80-afam.png" sizes="100vw" /></a>
    </span>
</span>
<p></p><h2 data-id="action-work-items-filtering">🛠️ <strong>Action Work Items Filtering</strong> </h2><p>Each action now has its own filter picker, using the same filter experience you know from Wrike’s standard filters. You can filter by item type, status, assignee, custom fields, importance, name, or any combination.</p><p>Filters are applied before the LLM runs, so filtered-out items do not consume credits.</p><p>This solves cases where a triaging agent on a project folder was acting on every sub-item (projects, milestones, everything). Now you can add a filter for “tasks only” and it ignores the rest. Or get more specific with “tasks in status New, assigned to nobody.” Each action in a multi-action agent can have different filters, so one action targets open tasks while another targets completed ones.</p><h2 data-id="new-custom-field-type-checkboxes"><strong>☑️ New Custom Field Type: Checkboxes </strong></h2><p>Agents now read and write checkbox custom fields.</p><p>Example use case: A “Reconciliation Complete” checkbox. Instead of someone manually checking the box after filling in all reporting fields, an agent can read those fields, decide whether the data is complete, and then check it, applying judgment, not just flipping a toggle.</p><p>The same pattern works for “Triaged,” “Assets Received,” “NDA Signed,” “Brief Approved,” or any binary state that currently requires someone to look at the task and decide.</p><h2 data-id="action-naming"><strong>🔤 Action Naming </strong></h2><p>You can now name individual actions inside multi-action agents. Instead of “Action 1,” “Action 2,” and “Action 3” in your activity logs, you’ll see labels like “Set Priority,” “Route to Team,” or “Post Summary.”</p><p><strong>Why this matters:</strong> It’s a small change that makes a big difference when you’re debugging a five-action agent and trying to figure out which action misfired.</p><h2 data-id="formula-field-reading"><strong>📖 Formula Field Reading </strong></h2><p>Agents can now read formulas and calculated custom field values during reasoning. If you have a field that computes a risk score, aggregates sub-item progress, or mirrors a native field value, agents can see it and use it in their decisions.</p><p>This is read-only (formula fields are computed, not editable), but it lets agents reason about derived metrics without you having to duplicate the data into a regular field.</p><p><strong>Pro tip:</strong> If there’s a field type that agents can’t read yet, create a formula field that references it. The formula output becomes visible to the agent and works as a bridge for any unsupported field type. Point a formula at it, and the agent sees the value.</p><h2 data-id="under-the-hood"><strong>Under the Hood 👇</strong></h2><ul><li><strong>Hard limits on sub-item processing:</strong> When an agent is configured to run on "all sub-items", you now get a warning at appointment time showing how many items will be affected. This prevents accidental action spikes on large folders.<br /><br /></li><li><strong>Improved date handling:</strong> The LLM layer now correctly interprets relative dates, date ranges, and timezone-aware scheduling.</li></ul><h2 data-id="agent-of-the-month-intake-deduplication-agent"><strong>Agent of the Month: Intake Deduplication Agent 💯</strong></h2><p>Teams that receive incoming work requests such as support tickets, project intake forms, or creative briefs deal with duplicates constantly. The same request comes in twice, or a new request describes something that already exists under a different name. Catching these manually requires someone to remember what’s already in the pipeline or to search before acting. Most teams don’t.</p><p>Why automations can’t do this: An automation can check exact field matches like “if Client Name equals X, do Y.” But incoming requests rarely use the exact same words. “Acme Corp” might come in as “Acme Corporation,” “ACME,” or “the Acme project from last quarter.” Matching requires reading the description, understanding intent, and judging similarity. That’s LLM territory.</p><p><strong>The solution:</strong> A two-action agent that uses three of this month’s new features, cross-folder lookup, location change, and action filtering, to scan your existing projects, detect duplicates by meaning (not just keywords), and route new requests automatically.</p><p><strong>How it works:</strong></p><ol><li>A new request lands in your intake folder.<br /><br /></li><li>The agent reads the request, then scans a reference folder (for example “Active Projects”) for items with similar names, descriptions, or scope using fuzzy matching, not exact matches.<br /><br /></li><li><strong>Action 1</strong> posts a comment: duplicate found, no match, or uncertain.<br /><br /></li><li><strong>Action 2</strong> routes the request to Active Projects if it’s new, or leaves it in intake for manual review if a duplicate was flagged.</li></ol><p>The pattern adapts to any domain: support tickets against known issues, bug reports against existing bugs, RFPs against past proposals. New item arrives → the agent scans a reference folder → judges similarity → then routes or flags.</p><p>Click the link below for the complete configuration, prompts, and customization guide.</p><p>👉 <a href="https://community.wrike.com/discussion/2845/intake-deduplication-agent" rel="nofollow noopener ugc"><strong>Solution and configuration</strong></a></p><h2 data-id="combining-this-month-s-features"><strong>Combining This Month's Features</strong></h2><p>Location change plus action filtering is a powerful combination. Imagine a project intake folder where new requests land:</p><ul><li><strong>Action 1</strong> (filtered to tasks in status “New” with no assignee): The agent reads the request, classifies the project type, and moves it to the correct team folder using location chips.<br /><br /></li><li><strong>Action 2</strong> (filtered to tasks in status “New”): The agent sets the due date based on the project type and the urgency described in the brief.<br /><br /></li><li><strong>Action 3:</strong> The agent posts a summary comment tagging the team lead (list the team leads and their folders in the general instruction, since the agent has no access to user data on its own).</li></ul><p>You end up with three named actions, “Route Request,” “Set Timeline,” and “Notify Team,” each with its own filter. The activity log shows exactly which action did what.</p><h2 data-id="questions"><strong>Questions?</strong></h2><p><strong>Want help setting up any of these features? </strong>Reach out to us in the comments.</p><p><strong>Need a different agent type?</strong> Contact your Customer Success Manager.</p><p><strong>Found a bug or have feedback?</strong> Use the feedback link in the product or let us know in the comments.</p><h2 data-id="on-our-radar"><strong>On Our Radar</strong></h2><p>Areas we're actively exploring:</p><ul><li><strong>Context filtering:</strong> Expanding what agents can see beyond the 3-level hierarchy limit.</li><li><strong>Sub-item trigger scope:</strong> Agents that trigger on changes to sub-items of a specific parent.</li><li><strong>Approval flow actions:</strong> Agents that start approval workflows and manage approvers.</li><li><strong>Sequential action dependencies:</strong> Actions where Step B uses the result of Step A.</li></ul><p>Your feedback shapes our roadmap. Keep testing, keep sharing what works (and what doesn't).</p>]]>
        </description>
    </item>
    <item>
        <title>Intake Deduplication Agent</title>
        <link>https://community.wrike.com/discussion/2845/intake-deduplication-agent</link>
        <pubDate>Thu, 26 Mar 2026 11:33:32 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2845@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This multi-action Wrike AI agent checks new intake requests for duplicates or near-duplicates by comparing their title and description against items in an “Active Projects” folder using meaning-based matching, not just exact text. It then posts a comment indicating whether a likely duplicate, no match, or an uncertain match was found, and automatically routes only unique requests into Active Projects while leaving possible duplicates in intake for manual review. It’s ideal for teams with high request volume, helps prevent duplicate work from entering the pipeline, and can be customized for different matching rules, routing logic, and follow-up actions like status updates.</p><p></p><p>Hello Community! 👋</p><p>Today we’re sharing a guide to set up an <strong>Intake Deduplication Agent</strong> that scans new incoming requests, compares them against items in your folder using meaning-based matching, and helps prevent duplicate work from entering your pipeline.</p><p>Below, you’ll find the full configuration, prompts, setup instructions, and customization ideas so you can plug this into your intake workflow ✔️</p><h2 data-id="agent-goal-use-case"><strong>Agent Goal / Use Case</strong></h2><p>Your intake folder receives more than 50 requests each week, and about 15% are duplicates or near-duplicates of work already in your pipeline. They are often the same project described in different words. By the time someone notices, a kickoff meeting may already be scheduled for work that already exists.</p><p>A standard automation rule can match exact field values, but it cannot recognize that “Acme Corp website redesign” and “ACME - redesign of corporate site” refer to the same project. </p><h2 data-id="how-it-works"><strong>How It Works</strong></h2><p>The <strong>Intake Deduplication Agent</strong> is a multi-action agent that:</p><ol><li>When a new request lands in your intake folder.<br /></li><li>Reads the new request’s title and description.<br /></li><li>It scans a reference folder (e.g. "Active Projects") for items with similar names, descriptions, or scope, fuzzy matching by meaning.<br /></li><li><strong>Action 1</strong> posts a comment: duplicate found, no match, or uncertain.<br /></li><li><strong>Action 2</strong> routes the request to Active Projects if it's new, or leaves it in intake for manual review if a duplicate was flagged.</li></ol><h2 data-id="features-used"><strong>Features Used</strong></h2><ul><li><strong>Cross-folder lookup</strong> — agent reads items from a different folder than where it's appointed<br /></li><li><strong>Location change</strong> — agent moves/adds items to the correct folder<br /></li><li><strong>Action filtering</strong> — actions only apply to tasks in Status = New<br /></li><li><strong>Action naming</strong> — actions are labeled "<strong>Check for Duplicates</strong>" and "<strong>Route Request</strong>" in the activity log</li></ul><h2 data-id="prerequisites"><strong>Prerequisites</strong></h2><p>Before setting this up, make sure you have:</p><h3 data-id="1-an-intake-folder"><strong>1. An Intake Folder</strong></h3><p>This is where new requests are created.</p><h3 data-id="2-an-active-projects-folder"><strong>2. An Active Projects Folder</strong></h3><p>This folder will act as the reference source the agent checks against.</p><h3 data-id="3-a-clear-request-structure"><strong>3. A clear request structure</strong></h3><p>At minimum, incoming items should have:</p><ul><li>Title</li><li>Description</li></ul><p>The better the request details, the better the duplicate detection quality.</p><div><table><colgroup><col /><col /><col /></colgroup><tr><th><p><strong>Setting</strong></p></th><th><p><strong>Value</strong></p></th></tr><tr><td><p><strong>Name</strong></p></td><td><p>Intake Deduplication Agent</p></td></tr><tr><td><p><strong>Trigger</strong></p></td><td><p>New item created</p></td></tr><tr><td><p><strong>Action Scope</strong></p></td><td><p>The work item where the trigger happened</p></td></tr><tr><td><p><strong>Acton 1 Name</strong></p></td><td><p>Check for Duplicates</p></td></tr><tr><td><p><strong>Action 1 Type</strong></p></td><td><p>Post a comment</p></td></tr><tr><td><p><strong>Action 1 Filter</strong></p></td><td><p>Item type: Task, Status: New</p></td></tr><tr><td><p><strong>Action 2 Name</strong></p></td><td><p>Route Request</p></td></tr><tr><td><p><strong>Action 2 Type</strong></p></td><td><p>Location change - Add to location</p></td></tr><tr><td><p><strong>Action 2 Filter</strong></p></td><td><p>Item type: Task, Status: New</p></td></tr><tr><td><p><strong>Appointment</strong></p></td><td><p>Your intake/requests folder</p></td></tr></table></div><h2 data-id="prompt"><strong>Prompt</strong></h2><h3 data-id="general-instructions"><strong>General Instructions</strong></h3><div><div><p>ROLE: You are an Intake Deduplication Specialist.</p><p><br /></p><p>CONTEXT: You have access to items in the current folder (incoming requests)</p><p>and in the "Active Projects" folder [📁 embed location chip for your Active Projects folder].</p><p><br /></p><p>REFERENCE DATA:</p><p>When comparing items, consider a match if ANY of these are similar:</p><p>- Client or company name (including abbreviations, variations, and partial matches)</p><p>- Project description or scope of work</p><p>- Key deliverables mentioned</p><p><br /></p><p>A match does not need to be exact. Use your judgment:</p><p>- "Acme Corp website redesign" and "ACME - redesign of corporate site" → MATCH (same client, same project)</p><p>- "Acme Corp website redesign" and "Acme Corp billing integration" → NOT A MATCH (same client, different project)<br />
- "Q3 brand refresh" and "Brand refresh - third quarter" → MATCH (same initiative, different wording)</p></div></div><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/J218N5UWEX1W\/gi-gimp.png&quot;,&quot;name&quot;:&quot;GI GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:114786,&quot;width&quot;:699,&quot;height&quot;:677,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FJ218N5UWEX1W%2Fgi-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3669,&quot;dateInserted&quot;:&quot;2026-03-26T11:24:03+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FJ218N5UWEX1W%2Fgi-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/J218N5UWEX1W/gi-gimp.png" alt="GI GIMP.png" height="677" width="699" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/J218N5UWEX1W/gi-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/J218N5UWEX1W/gi-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/J218N5UWEX1W/gi-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/J218N5UWEX1W/gi-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/J218N5UWEX1W/gi-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/J218N5UWEX1W/gi-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/J218N5UWEX1W/gi-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h3 data-id="action-1-post-a-comment"><strong><br />
Action 1 — Post a Comment</strong></h3><p><strong>Action name</strong>: Check for Duplicates</p><p>Copy-paste the following into <strong>Action 1:</strong></p><div><div><p>Read the title and description of this incoming request.</p><p><br /></p><p>Scan all items in the "Active Projects" folder [📁 location chip].</p><p><br /></p><p>Compare each active project against this request using the matching</p><p>criteria in the general instruction.</p><p><br /></p><p>IF a likely match is found:</p><p>Post a comment:</p><p>"⚠️ Possible duplicate detected.</p><p>This request appears to match: [matched item name].</p><p>Similarity: [brief explanation of what matched — e.g. same client name,</p><p>overlapping deliverables, similar project scope].</p><p>Please review before proceeding."</p><p><br /></p><p>IF no match is found:</p><p>Post a comment:</p><p>"✅ No duplicates found in Active Projects. Safe to proceed."</p><p><br /></p><p>IF you are uncertain whether two items match:</p><p>Post a comment:</p><p>"⚠️ Uncertain match: [item name].</p><p>[reason for uncertainty — e.g. same client but unclear if same project].</p><p>Manual review recommended."</p><p><br /></p><p>Do NOT guess. When in doubt, flag it as uncertain rather than declaring</p><p>a match or no-match.</p></div></div><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/I5S1RP0P6JSY\/action-1-gimp.png&quot;,&quot;name&quot;:&quot;Action 1 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:51842,&quot;width&quot;:703,&quot;height&quot;:322,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FI5S1RP0P6JSY%2Faction-1-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3670,&quot;dateInserted&quot;:&quot;2026-03-26T11:25:14+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FI5S1RP0P6JSY%2Faction-1-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png" alt="Action 1 GIMP.png" height="322" width="703" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/I5S1RP0P6JSY/action-1-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h3 data-id="action-2-location-change-add-to-location"><strong><br />
Action 2 — Location Change (Add to location)</strong></h3><p><strong>Action name:</strong> Route Request</p><p>Copy-paste the following into <strong>Action 2:</strong></p><div><div><p>Read the comment posted by the previous action on this item.<br /><br /></p><p>IF the comment says "No duplicates found" (the request is new and unique):</p><p>Add this item to the "Active Projects" folder [📁 location chip].</p><p><br /></p><p>IF the comment says "Possible duplicate detected" or "Uncertain match":</p><p>Do NOT move this item. Leave it in the intake folder for manual review.</p></div></div><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/9FJQ5SIGYD72\/action-2-gimp.png&quot;,&quot;name&quot;:&quot;Action 2 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:51801,&quot;width&quot;:662,&quot;height&quot;:392,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F9FJQ5SIGYD72%2Faction-2-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3671,&quot;dateInserted&quot;:&quot;2026-03-26T11:26:29+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F9FJQ5SIGYD72%2Faction-2-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png" alt="Action 2 GIMP.png" height="392" width="662" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/9FJQ5SIGYD72/action-2-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h2 data-id="setup-steps"><br /><strong>Setup Steps</strong></h2><h3 data-id="step-1-create-the-agent"><strong>Step 1: Create the Agent</strong></h3><ul><li>Go to Space Settings → AI Agents → Create Custom AI Agent</li><li>Name the agent “<strong>Intake Deduplication Agent</strong>”</li><li><strong>Set the Trigger</strong> to “New item created”</li><li><strong>Paste the General Instructions</strong> into the instructions field. Replace the location chip placeholder with your actual "Active Projects" folder — type @ or use the folder picker to embed the chip .</li></ul><h3 data-id="step-2-configure-action-1"><strong>Step 2: Configure Action 1</strong></h3><ul><li>Add Action 1</li><li>Select Post a comment</li><li>Name it “<strong>Check for Duplicates”</strong></li><li>Paste the Action 1 prompt</li><li>Add the filter: Item type = Task, Status = New</li></ul><h3 data-id="step-3-configure-action-2"><strong>Step 3: Configure Action 2</strong></h3><ul><li>Add Action 2</li><li>Select Location change (Add to location)</li><li>Name it “<strong>Route Request</strong>”</li><li>Paste the Action 2 prompt </li><li>Select your "Active Projects" folder as the target location</li><li>Add the same filter: Item type = Task, Status = New</li></ul><h3 data-id="step-4-appoint-the-agent"><strong>Step 4: Appoint the Agent</strong></h3><ul><li>Appoint the agent to your intake/requests folder.</li></ul><h3 data-id="step-5-test-in-playground"><strong>Step 5: Test in Playground</strong></h3><ul><li>Before rolling it out broadly, test with a few examples:</li></ul><h2 data-id="customization"><strong>Customization </strong></h2><h3 data-id="1-different-matching-criteria"><strong>1. Different matching criteria</strong></h3><p>Adjust the "REFERENCE DATA" section in the general instruction to match your domain:</p><p>Examples:</p><ul><li><strong>Support tickets</strong>: Match on error messages, affected product/module, and customer name<br /></li><li><strong>Bug reports:</strong> Match on reproduction steps, error codes, and affected component<br /></li><li><strong>RFPs / proposals:</strong> Match on client name, industry, and requested services<br /></li><li><strong>Creative briefs:</strong> Match on campaign name, brand, and target audience</li></ul><h3 data-id="2-stricter-or-looser-matching"><strong>2. Stricter or looser matching</strong></h3><ul><li>To reduce false positives: Add "Only flag a match if at least two criteria match simultaneously."<br /></li><li>To catch more duplicates: Add "Also consider partial matches , if the client name matches but the project description is different, flag as uncertain."</li></ul><h3 data-id="3-different-routing-behavior"><strong>3. Different routing behavior</strong></h3><ul><li><strong>Move instead of add:</strong> Change Action 2 from "Add to location" to "Move to location" if you want the item removed from intake entirely.<br /></li><li><strong>Route to different folders based on type:</strong> Add more location chips and adjust the Action 2 prompt: "If no duplicate and the request mentions 'design', add to the Creative Projects folder. If it mentions 'engineering', add it to the Engineering Backlog folder."</li></ul><h3 data-id="4-adding-a-status-change">4<strong>. Adding a status change</strong></h3><p>If you want the agent to also update the request's status after routing, add a third action:</p><ul><li><strong>Action 3 name:</strong> "Update Status"</li><li><strong>Action 3 type:</strong> Change status</li><li><strong>Prompt:</strong> "If this item was moved to Active Projects, change its status to 'In Review.' If it was flagged as a duplicate, change its status to 'Needs Review.”</li></ul><h2 data-id="tips"><strong>Tips</strong></h2><ul><li><strong>Start small.</strong> Appoint to one intake folder with 10-20 items in the reference folder. Verify the matching quality before scaling.<br /></li><li><strong>Check the activity log.</strong> The agent's reasoning shows exactly why it flagged (or didn't flag) a match. Use action names "Check for Duplicates" and "Route Request" to quickly find what you're looking for.<br /></li><li><strong>Reference folder size matters.</strong> Cross-folder lookup works well up to a few hundred items. If your Active Projects folder has thousands of items, consider narrowing with a filter or using a more specific reference folder.<br /></li><li><strong>The uncertain category is your friend.</strong> It's better to flag an uncertain match for human review than to miss a duplicate or incorrectly block a new request.</li></ul>]]>
        </description>
    </item>
    <item>
        <title>Help Us Shape Our Next AI Agents Live Session 🙋</title>
        <link>https://community.wrike.com/discussion/2858/help-us-shape-our-next-ai-agents-live-session</link>
        <pubDate>Wed, 01 Apr 2026 08:49:21 +0000</pubDate>
        <category>Wrike AI</category>
        <dc:creator>Elizaveta Bogdanova</dc:creator>
        <guid isPermaLink="false">2858@/discussions</guid>
        <description><![CDATA[<p></p><p>Hi everyone! 👋</p><p>Exciting news - we’re currently preparing an <strong>upcoming session on Wrike AI Agents</strong> (date coming soon!), and we’d love your input while we’re still in prep mode! </p><p>If you could automate <em>one</em> part of your work with AI Agents, what would it be?</p><p>Which of these would be most valuable for you to see in action:</p><ul><li>Turning messy requests into clear, structured tasks (intake/triage)</li><li>Analyzing project health using multiple signals</li><li>Building a project plan from a simple goal</li><li>Summarizing work and sharing updates automatically</li><li>Something else? Tell us in the comments 👇</li></ul><p><strong>Drop your vote or share your use case below</strong> - we’ll use your input to shape the session 💬 </p><p>Stay tuned - we’ll be announcing the date soon! 🚀</p>]]>
        </description>
    </item>
    <item>
        <title>Stale Task Escalation Agent</title>
        <link>https://community.wrike.com/discussion/2736/stale-task-escalation-agent</link>
        <pubDate>Wed, 25 Feb 2026 11:37:00 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2736@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This multi-action Wrike AI agent automatically tracks how many times a task is sent back for review by incrementing a numeric “Revision Count” custom field and posting a short comment each time a task enters a selected review status. It gives you dashboard-ready data on revision cycles, replaces manual tracking or external integrations, and is fully customizable to your workflow (field name, trigger status, and comment format).</p><p>Hello Community! 👋</p><p>We’re thrilled to share a <strong>Stale Task Escalation Agent (agent chaining pattern) </strong>that automatically detects stale work, escalates it, and notifies the right person, all natively in Wrike using multiple agents that trigger each other ⚙️</p><p>Below, you’ll find the complete configuration, prompts, setup instructions, and customization tips so you can adapt this pattern to your own SLA/escalation workflows ✔️</p><h2 data-id="agent-goal-use-case"><strong>Agent Goal / Use Case</strong></h2><p>Automatically detect tasks that have been “stuck” too long in certain statuses, escalate them by changing the task status, and notify a project manager with a comment without manual monitoring or external tools.</p><p><strong>Typical scenario:</strong></p><p>Approvals, reviews, or blocked tasks sit for days (or weeks) before anyone notices. By the time someone realizes, it’s urgent and requires a fire drill.</p><p><strong>Example:</strong> “Legal reviews can sit for weeks. We need automated escalation when approval time exceeds our SLA.”</p><p>This agent chain solves that by:</p><ul><li>Scanning for stuck tasks on a schedule.</li><li>Flagging stale tasks via a custom field.</li><li>Escalating their status automatically.</li><li>Notifying the project manager via comment.</li></ul><p><strong>Time Saved:</strong> ~3–5 hours/week per manager (no more manual dashboard checks or chasing overdue items)</p><h2 data-id="how-it-works"><strong>How It Works</strong></h2><p>This pattern uses <strong>three simple agents chained together</strong>:</p><ul><li><strong>Agent 1 (Stale Detector) -</strong> Scheduled daily, finds stuck tasks, sets “Stale” field to “Yes”.</li><li><strong>Agent 2 (Escalator) -</strong> Triggered by field change, changes status to “Escalated”.</li><li><strong>Agent 3 (Notifier) -</strong> Triggered by status change, posts comment @mentioning project manager.</li></ul><p><strong>Why three agents?</strong></p><ul><li>Each agent does one thing (detect → escalate → notify).</li><li>Easier to test and debug each step.</li><li>Highly composable: you can swap out detectors, escalation rules, or notification logic independently.</li></ul><h2 data-id="prerequisites"><strong>Prerequisites</strong></h2><p>Before setting up the agents, you’ll need:</p><p><strong>Custom Field: “Stale”</strong></p><ul><li>Type: Dropdown (single-select)</li><li>Values: "Yes", "No" (or "Yes" plus blank as default)</li><li>Make sure it’s added and visible in the space/folder/project where the agents will run</li></ul><p><strong>Workflow Status: “Escalated” (or similar)</strong></p><ul><li>Add an “Escalated” / “Needs Attention” type status to your task workflow</li><li>This is the status that indicates a task has been formally escalated</li></ul><h2 data-id="agent-1-stale-detector"><strong>Agent 1: Stale Detector</strong></h2><p>This agent runs daily, scans for stuck tasks, and marks them as stale.</p><p><strong>Configuration:</strong></p><div><table><colgroup><col /><col /><col /></colgroup><tr><th><p><strong>Setting</strong></p></th><th><p><strong>Value</strong></p></th></tr><tr><td><p><strong>Name</strong></p></td><td><p>Stale Detector</p></td></tr><tr><td><p><strong>Scope</strong></p></td><td><p>Any subitem of the item where the agent was added</p></td></tr><tr><td><p><strong>Trigger</strong></p></td><td><p>Scheduled - Daily at 9:00 AM</p></td></tr><tr><td><p><strong>Action</strong></p></td><td><p>Change custom field</p></td></tr></table></div><p><strong>Prompt:</strong></p><p><strong>General instructions:</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/3FJBFSBRI50J\/agent-1-1-gimp.png&quot;,&quot;name&quot;:&quot;Agent 1 - 1 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:144658,&quot;width&quot;:1584,&quot;height&quot;:1110,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F3FJBFSBRI50J%2Fagent-1-1-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3568,&quot;dateInserted&quot;:&quot;2026-02-25T11:01:00+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F3FJBFSBRI50J%2Fagent-1-1-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png" alt="Agent 1 - 1 GIMP.png" height="1110" width="1584" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/3FJBFSBRI50J/agent-1-1-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>You are a Stale Task Detector.</p><p><br /></p><p>Mark tasks as stale when they've been stuck too long.</p><p><br /></p><p>=== RULES ===</p><p><br /></p><p>Check tasks in "Pending Approval" status.</p><p><br /></p><p>If in that status for more than 5 days → mark as stale.</p></div></div><p><strong>Action:</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/C9945BJADSV9\/agent-1-2-gimp.png&quot;,&quot;name&quot;:&quot;Agent 1 - 2 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:130099,&quot;width&quot;:1394,&quot;height&quot;:1132,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FC9945BJADSV9%2Fagent-1-2-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3569,&quot;dateInserted&quot;:&quot;2026-02-25T11:01:26+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FC9945BJADSV9%2Fagent-1-2-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png" alt="Agent 1 - 2 GIMP.png" height="1132" width="1394" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/C9945BJADSV9/agent-1-2-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>=== ACTION ===</p><p><br /></p><p>Set "Stale" field to "Yes".</p><p><br /></p><p>=== SKIP ===</p><p><br /></p><p>- Tasks where status changed in last 24 hours</p><p>- Tasks already marked Stale = Yes</p></div></div><h2 data-id="agent-2-escalator"><strong>Agent 2: Escalator</strong></h2><p>This agent responds when a task is marked stale and changes its status.</p><div><table><colgroup><col /><col /><col /></colgroup><tr><th><p><strong>Setting</strong></p></th><th><p><strong>Value</strong></p></th></tr><tr><td><p><strong>Name</strong></p></td><td><p>Stale Escalator</p></td></tr><tr><td><p><strong>Scope</strong></p></td><td><p>Any subitem of the item where the agent was added</p></td></tr><tr><td><p><strong>Trigger</strong></p></td><td><p>Custom field changed - “Stale”</p></td></tr><tr><td><p><strong>Action</strong></p></td><td><p>Change status</p></td></tr></table></div><p><strong>Prompt:</strong></p><p><strong>General instructions:</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/H8ZLKBZFMWGB\/agent-2-1-gimp.png&quot;,&quot;name&quot;:&quot;Agent 2 -1 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:147027,&quot;width&quot;:1376,&quot;height&quot;:1172,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FH8ZLKBZFMWGB%2Fagent-2-1-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3570,&quot;dateInserted&quot;:&quot;2026-02-25T11:03:34+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FH8ZLKBZFMWGB%2Fagent-2-1-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png" alt="Agent 2 -1 GIMP.png" height="1172" width="1376" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/H8ZLKBZFMWGB/agent-2-1-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>You are a Stale Task Escalator.</p><p></p><p></p><p>When "Stale" field changes to "Yes" and status is "Pending Approval", change status to "Escalated".</p><p><br /></p><p>=== SKIP ===</p><p><br /></p><p>Don't change status if task is already "Completed", "Cancelled", or "Escalated"</p></div></div><p><strong>Action:</strong></p><p>Change Status</p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/T5N9MQU8VABC\/agent-2-2-gimp.png&quot;,&quot;name&quot;:&quot;Agent 2 - 2 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:117752,&quot;width&quot;:1410,&quot;height&quot;:1012,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FT5N9MQU8VABC%2Fagent-2-2-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3571,&quot;dateInserted&quot;:&quot;2026-02-25T11:04:39+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FT5N9MQU8VABC%2Fagent-2-2-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png" alt="Agent 2 - 2 GIMP.png" height="1012" width="1410" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/T5N9MQU8VABC/agent-2-2-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h2 data-id="agent-3-escalation-notifier"><strong>Agent 3: Escalation Notifier</strong></h2><p>This agent posts a comment when a task is escalated, notifying the project manager.</p><div><table><colgroup><col /><col /><col /></colgroup><tr><th><p><strong>Setting</strong></p></th><th><p><strong>Value</strong></p></th></tr><tr><td><p><strong>Name</strong></p></td><td><p>Escalation Notifier</p></td></tr><tr><td><p><strong>Scope</strong></p></td><td><p>Any subitem of the item where the agent was added</p></td></tr><tr><td><p><strong>Trigger</strong></p></td><td><p>Status changed - to “Escalated”</p></td></tr><tr><td><p><strong>Action</strong></p></td><td><p>Post comment</p></td></tr></table></div><p><strong>Prompt:</strong></p><p><strong>General instructions:</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/C718H0INQWIJ\/agent-3-1-gimp.png&quot;,&quot;name&quot;:&quot;Agent 3 -1 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:94610,&quot;width&quot;:1254,&quot;height&quot;:1050,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FC718H0INQWIJ%2Fagent-3-1-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3572,&quot;dateInserted&quot;:&quot;2026-02-25T11:08:16+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FC718H0INQWIJ%2Fagent-3-1-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png" alt="Agent 3 -1 GIMP.png" height="1050" width="1254" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/C718H0INQWIJ/agent-3-1-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>You are an Escalation Notifier.</p><p><br /></p><p>When status changes to "Escalated", post a comment to notify the project manager.</p></div></div><p><strong>Action:</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/2YJCR89X8EOP\/agent-3-2-gimp.png&quot;,&quot;name&quot;:&quot;Agent 3 - 2 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:124683,&quot;width&quot;:1246,&quot;height&quot;:980,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F2YJCR89X8EOP%2Fagent-3-2-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3573,&quot;dateInserted&quot;:&quot;2026-02-25T11:09:45+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F2YJCR89X8EOP%2Fagent-3-2-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png" alt="Agent 3 - 2 GIMP.png" height="980" width="1246" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/2YJCR89X8EOP/agent-3-2-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>=== ACTION ===<br /></p><p>Post comment:</p><p>"⚠️ This task has been escalated due to extended time in approval. Please review and help unblock."</p></div></div><p><strong>Note:</strong> Configure the @mention in the comment action parameters when setting up the agent. Select the project manager to notify.</p><h2 data-id="the-complete-chain"><strong>The Complete Chain</strong></h2><div><div><p>Agent 1: Stale Detector (Scheduled daily)<br /></p><p>Action: Sets "Stale" field to "Yes"</p></div></div><p> <strong>Triggers 👇</strong></p><div><div><p>Agent 2: Escalator (Field change trigger)<br /><br />
Action: changes status to "Escalated"</p></div></div><p><strong>Triggers 👇</strong></p><div><div><p>Agent 3: Notifier (Status change trigger)<br /></p><p>Action: Posts comment @mentioning PM</p></div></div><h2 data-id="step-by-step-setup"><strong>Step-by-Step Setup </strong></h2><h3 data-id="step-1-create-the-custom-field"><strong>Step 1: Create the Custom Field</strong></h3><ul><li>Go to Space Settings → Custom Fields</li><li>Create new dropdown (Single select) field named “Stale”</li><li>Add values: “Yes”, “No”</li><li>Save</li></ul><h3 data-id="step-2-add-escalation-status-if-needed"><strong>Step 2: Add Escalation Status (if needed)</strong></h3><ul><li>Go to Space Settings → Workflows</li><li>Edit your task workflow</li><li>Add status “Escalated” (or similar)</li><li>Save</li></ul><h3 data-id="step-3-deploy-agent-1-stale-detector"><strong>Step 3: Deploy Agent 1 (Stale Detector)</strong></h3><ul><li>Go to Space Settings → AI Agents</li><li>Click “+ Custom AI agent”</li></ul><p>Configure:</p><ul><li><strong>Name:</strong> Stale Detector</li><li><strong>Scope:</strong> Any subitem of the item where the agent was added</li><li><strong>Trigger:</strong> Scheduled → Daily → 9:00 AM</li><li><strong>Action:</strong> Change custom field → “Stale”<br /></li><li>Paste and customize the prompt</li><li>Test in Playground with sample tasks</li><li>Appoint to your target folder/project</li></ul><h3 data-id="step-4-deploy-agent-2-escalator"><strong>Step 4: Deploy Agent 2 (Escalator)</strong></h3><ul><li>Click “+ Custom AI agent” again</li></ul><p>Configure:</p><ul><li><strong>Name:</strong> Stale Escalator</li><li><strong>Scope:</strong> Any subitem of the item where the agent was added</li><li><strong>Trigger:</strong> Custom field changed → “Stale”</li><li><strong>Action:</strong> Change status<br /></li><li>Paste and customize the prompt</li><li>Test in Playground</li><li>Appoint to the SAME folder/project as Agent 1</li></ul><h3 data-id="step-5-deploy-agent-3-notifier"><strong>Step 5: Deploy Agent 3 (Notifier)</strong></h3><ul><li>Click “+ Custom AI agent” again</li></ul><p>Configure:</p><ul><li><strong>Name:</strong> Escalation Notifier</li><li><strong>Scope:</strong> Any subitem of the item where the agent was added</li><li><strong>Trigger:</strong> Status changed → “Escalated”</li><li><strong>Action:</strong> Post comment<br /></li><li>Paste and customize the prompt (update project manager name)</li><li>Test in Playground</li><li>Appoint to the SAME folder/project as Agents 1 and 2</li></ul><h3 data-id="step-6-verify-the-chain"><strong>Step 6: Verify the Chain</strong></h3><ul><li>Find a task that should be stale (or create a test task)</li><li>Manually set it to “Pending Approval” status with an old date</li><li>Wait for Agent 1’s scheduled run (or trigger manually in Playground)</li><li>Verify “Stale” field changes to “Yes”</li><li>Verify Agent 2 triggers and changes status to “Escalated”</li><li>Verify Agent 3 triggers and posts comment @mentioning project manager </li></ul><h2 data-id="customization-examples"><strong>Customization Examples</strong></h2><h3 data-id="for-approval-workflows"><strong>For Approval Workflows</strong></h3><div><div><p>Monitor these statuses:</p><p>- "Pending Approval" - stale after 5 days</p><p>- "Legal Review" - stale after 7 days</p><p>- "Executive Sign-off" - stale after 3 days</p><p><br /></p><p>Escalation status: "Approval Overdue"</p></div></div><h3 data-id="for-creative-agency-teams"><strong>For Creative/Agency Teams</strong></h3><div><div><p>Monitor these statuses:</p><p>- "Client Review" - stale after 5 days</p><p>- "Waiting for Assets" - stale after 7 days</p><p>- "Pending Feedback" - stale after 3 days</p><p><br /></p><p>Escalation status: "At Risk"</p></div></div><h3 data-id="for-it-support-teams"><strong>For IT/Support Teams</strong></h3><div><div><p>Monitor these statuses:</p><p>- "Waiting on Vendor" - stale after 10 days</p><p>- "Pending Info" - stale after 3 days</p><p>- "On Hold" - stale after 14 days</p><p><br /></p><p>Escalation status: "Needs Attention"</p></div></div><h3 data-id="expected-results"><strong>Expected Results</strong></h3><p><strong>Immediate Impact:</strong><br />
- Stuck tasks get flagged automatically<br />
- No items slip through the cracks<br />
- Consistent SLA enforcement</p><p><strong>Team Benefits:</strong><br />
- Managers don’t waste time hunting for overdue items<br />
- “Escalated” status creates dashboard visibility<br />
- Clear signal that something needs attention</p><p><strong>Organizational Benefits:</strong><br />
- Faster resolution of stuck work<br />
- Data on where bottlenecks occur (which statuses, how often)<br />
- Reduced fire drills from late discovery</p><h2 data-id="troubleshooting"><strong>Troubleshooting</strong></h2><p><strong>Agent 1 not marking tasks as stale:</strong><br />
- Verify status names match exactly (case-sensitive).<br />
- Check that tasks have been in status long enough.<br />
- Confirm agent can see tasks in the folder (permissions).</p><p><strong>Agent 2 not triggering:</strong><br />
- Verify it’s watching the “Stale” field specifically.<br />
- Check that both agents are appointed to the same location.<br />
- Confirm Agent 1 is actually changing the field (check Activity Dashboard).</p><p><strong>Agent 3 not commenting:</strong><br />
- Verify it’s watching for status change to “Escalated” specifically.<br />
- Check that all three agents are appointed to the same location.<br />
- Confirm Agent 2 is actually changing the status.</p><p><strong>Chain feels slow:</strong><br />
- Agent 2 triggers within seconds of Agent 1’s field change.<br />
- If Agent 1 is scheduled for 9 AM, the full chain completes by 9:01 AM.<br />
- For faster detection, increase Agent 1’s schedule frequency.</p><p><strong>Too many escalations:</strong><br />
- Increase the staleness thresholds.<br />
- Add more exclusions (certain folders, certain statuses).<br />
- Check that the 24-hour grace period is working.</p>]]>
        </description>
    </item>
    <item>
        <title>Urgency Classifier Agent</title>
        <link>https://community.wrike.com/discussion/2822/urgency-classifier-agent</link>
        <pubDate>Tue, 17 Mar 2026 09:33:12 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2822@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This Wrike AI agent automatically reads new tasks, detects urgency signals in the title and description, and sets the Priority field (High / Normal / Low)  or asks the assignee to clarify when intent is ambiguous. It removes the need for manual triage while avoiding risky mislabeling of unclear requests.</p><p>Hello Community! 👋</p><p>Today we’re sharing a guide to set up an <strong>Urgency Classifier Agent</strong> that scans new tasks, looks for urgency cues (like “ASAP,” “no rush,” or “emergency”), and automatically updates a <strong>Priority</strong> custom field or posts a comment asking for clarification when it can’t confidently decide.</p><p>Below, you’ll find the full configuration, prompt, setup instructions, and customization ideas so you can plug this into your intake or backlog workflows ✔️</p><h2 data-id="agent-goal-use-case"><strong>Agent Goal / Use Case</strong></h2><p>Teams often waste time manually reading every incoming task to decide if it’s <strong>High</strong>, <strong>Normal</strong>, or <strong>Low</strong> priority. Some tasks say “urgent” or “ASAP,” others say “no rush,” and many have no clear signal at all.</p><h2 data-id="how-it-works"><strong>How It Works</strong></h2><p>The <strong>Urgency Classifier Agent</strong> is a single multi‑action agent that:</p><ul><li>Reads the task Title and Description.</li><li>Detects urgency phrases (e.g., “ASAP,” “urgent,” “no rush”).</li><li>Automatically sets the Priority custom field to High / Normal / Low for clear cases.</li><li>If urgency is unclear, it leaves Priority blank and posts a comment asking the assignee to clarify.</li></ul><p><strong>Technical flow:</strong></p><ul><li><strong>Trigger:</strong> When an item is created, or when its Description changes.</li><li><strong>Action 1:</strong> Update the Priority custom field based on urgency cues.</li><li><strong>Action 2 (conditional):</strong> If urgency is ambiguous, post a comment asking the assignee to set the Priority.</li></ul><p>This provides automated triage for obvious cases and a safe fallback (human clarification) when intent isn’t clear.</p><h2 data-id="prerequisites"><strong>Prerequisites</strong></h2><ul><li><strong>Custom Field:</strong> Priority</li><li><strong>Type:</strong> Dropdown (single-select)  </li><li><strong>Values:</strong> High, Normal, Low  </li><li><strong>Location:</strong> Add this field to the folder/project (or space) where the agent will run.</li></ul><h2 data-id="agent-urgency-classifier"><strong>Agent: Urgency Classifier</strong></h2><p>This agent analyzes the <strong>Task Title</strong> and <strong>Description</strong>, sets <strong>Priority</strong> when clear, and posts a clarification comment when not.</p><h3 data-id="configuration"><strong>Configuration:</strong></h3><div><table><colgroup><col /><col /><col /></colgroup><tr><th scope="col"><p><strong>Setting</strong></p></th><th scope="col"><p><strong>Value</strong></p></th></tr><tr><td><p><strong>Name</strong></p></td><td><p>Urgency Classifier</p></td></tr><tr><td><p><strong>Scope</strong></p></td><td><p>[Applies to subitems under that location (so all tasks in that folder/project are covered)]</p></td></tr><tr><td><p><strong>Trigger</strong></p></td><td><p>Either “Item created” or “Description changed”, depending on what best fits your workflow</p></td></tr></table></div><p><strong>Please Note:</strong> Each agent can have one trigger configured at a time. Wrike’s AI Agent offers multiple trigger types; however, the following two are typically the most logical for this use case:</p><ul><li><strong>New item created</strong> – use this if you want the agent to run as soon as a new task is created.</li><li><strong>Description changed</strong> – use this if you want the agent to run when someone updates the task description.</li></ul><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/890JCOM5RTBG\/trigger-explaination-gimp.png&quot;,&quot;name&quot;:&quot;Trigger explaination GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:79621,&quot;width&quot;:1422,&quot;height&quot;:556,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F890JCOM5RTBG%2Ftrigger-explaination-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3642,&quot;dateInserted&quot;:&quot;2026-03-17T09:27:02+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F890JCOM5RTBG%2Ftrigger-explaination-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png" alt="Trigger explaination GIMP.png" height="556" width="1422" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/890JCOM5RTBG/trigger-explaination-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h2 data-id="prompt"><strong>Prompt </strong></h2><h3 data-id="general-instructions"><strong>General Instructions:</strong></h3><p>Copy-paste the following into the <strong>General Instructions</strong> field:</p><div><div><p><strong>Role:</strong> You are a Triage Assistant. Your role is to analyze the urgency of the current task and ensure it is prioritized correctly based on text-based cues.</p><p><br /></p><p><strong>Objective:</strong> Scan the task title and description to detect urgency and update the "Priority" custom field accordingly. If urgency is unclear, request clarification from the assignee.</p><p></p><p></p><p><strong>Context &amp; Scope</strong></p><p><br /></p><p>Read: Task Title and Task Description.</p><p>Ignore: External links or historical comments unless they contain specific instructions about priority.</p><p></p><p></p><p><strong>Logic Rules</strong></p><p><strong></strong></p><p></p><p><strong>High Priority:</strong> If the title or description contains keywords like "urgent," "ASAP," "emergency," "critical," or "SLA," set the field to High.</p><p><strong>Normal Priority:</strong> If the task appears to be a standard request or uses the word "Normal," set the field to Normal.</p><p><strong>Low Priority:</strong> If the task contains phrases like "low priority," "when possible," "no rush," or "backlog," set the field to Low.</p><p><strong>Edge Case (Clarification):</strong> If no keywords are found or the intent is ambiguous, leave the field blank and trigger the comment action.</p></div></div><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/MH6DFWUQ8ZII\/general-instructions-gimp.png&quot;,&quot;name&quot;:&quot;General Instructions GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:75890,&quot;width&quot;:787,&quot;height&quot;:396,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FMH6DFWUQ8ZII%2Fgeneral-instructions-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3643,&quot;dateInserted&quot;:&quot;2026-03-17T09:28:13+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FMH6DFWUQ8ZII%2Fgeneral-instructions-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png" alt="General Instructions GIMP.png" height="396" width="787" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/MH6DFWUQ8ZII/general-instructions-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h3 data-id="action-1-update-custom-field"><strong>Action 1: Update Custom Field</strong></h3><ul><li><strong>Custom Field:</strong> Priority</li><li><strong>Values:</strong> "High", "Normal", "Low"</li></ul><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/R3OW1JEHP13W\/action-1-gimp.png&quot;,&quot;name&quot;:&quot;Action 1 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:14726,&quot;width&quot;:409,&quot;height&quot;:159,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FR3OW1JEHP13W%2Faction-1-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3644,&quot;dateInserted&quot;:&quot;2026-03-17T09:28:43+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FR3OW1JEHP13W%2Faction-1-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png" alt="Action 1 GIMP.png" height="159" width="409" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/R3OW1JEHP13W/action-1-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h3 data-id="action-2-post-a-comment"><strong>Action 2: Post a Comment </strong></h3><div><div><p><strong>Condition:</strong> Only if Logic Rule #4 applies.</p><p><strong>Template:</strong> "@[Assignee], I was unable to automatically determine the priority for this task based on the current description. Could you please update the Priority field?"</p><p></p><p></p><p><strong>Reasoning:</strong> In the agent activity log, briefly state which keywords triggered your decision.</p><p><br /></p><p><strong>Examples</strong></p><p><br /></p><p>Input: Title: "Emergency Server Fix" → Action: Set Priority to "High"</p><p>Input: Title: "Update Logo" / Description: "We should look at this at some point." → Action: Set Priority to "Low"</p><p>Input: Title: "Monthly Report" / Description: "Please send the report." → Action: Leave field blank + Comment: "@[Assignee], I was unable to automatically determine the priority…"</p></div></div><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/4DIXXVX6YIHG\/action-2-gimp.png&quot;,&quot;name&quot;:&quot;Action 2 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:81326,&quot;width&quot;:739,&quot;height&quot;:499,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F4DIXXVX6YIHG%2Faction-2-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3645,&quot;dateInserted&quot;:&quot;2026-03-17T09:30:18+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F4DIXXVX6YIHG%2Faction-2-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png" alt="Action 2 GIMP.png" height="499" width="739" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/4DIXXVX6YIHG/action-2-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<h2 data-id="setup-steps"><strong>Setup Steps</strong></h2><h3 data-id="step-1-create-the-custom-field"><strong>Step 1: Create the Custom Field</strong></h3><p>1. Go to <strong>Space Settings → Custom Fields</strong>.  </p><p>2. Create a dropdown field called "<strong>Priority</strong>".  </p><p>3. Add values: High, Normal, Low.  </p><p>4. Add it to the relevant intake/backlog folder or project.</p><h3 data-id="step-2-create-the-agent"><strong>Step 2: Create the Agent</strong></h3><p>1. Go to <strong>Space Settings → AI Agents → Create Custom AI agent</strong>.  </p><p>2. Name it <strong>Urgency Classifier</strong>.  </p><p>3. <strong>Scope:</strong> Set the scope to “Any subitem of the item where the agent was added” so it applies to all subitems in that location, ensuring all tasks in that folder or project are covered.</p><p>4. <strong>Trigger</strong>: Choose one:</p><ul><li>Item created, if you want it to run when a new task is created, or</li><li>Description changed, if you want it to run when someone edits the description.</li></ul><p>5. Paste the <strong>General Instructions</strong> prompt above into the agent’s instructions.</p><h3 data-id="step-3-configure-actions"><strong>Step 3: Configure Actions</strong></h3><p>1. <strong>Action 1 – Update Custom Field  </strong></p><ul><li>Custom Field: Priority  </li><li>Values: High / Normal / Low  </li></ul><p>2. <strong>Action 2 – Post Comment </strong> </p><ul><li>Condition: Only when Logic Rule #4 applies (no clear urgency).  </li><li>Comment template:  @[Assignee], I was unable to automatically determine the priority for this task based on the current description. Could you please update the Priority field?</li></ul><h3 data-id="step-4-test-in-playground"><strong>Step 4: Test in Playground</strong></h3><p>Try sample tasks:</p><ul><li><strong>High:</strong> Title: “Emergency server fix” → Expect: Priority = High <br />
 </li><li><strong>Low: </strong>Description: “No rush on this, add to backlog when possible” → Expect: Priority = Low <br />
 </li><li><strong>Ambiguous: </strong>Title: “Monthly report” / Description: “Please Review” → Expect: Priority blank + clarification comment</li></ul><h3 data-id="step-5-deploy"><strong>Step 5: Deploy</strong></h3><ul><li>Add the agent to your intake folder or project where new tasks are created.</li></ul><h2 data-id="customization-tips"><strong>Customization Tips</strong></h2><ul><li><strong>Add your own keywords</strong>: Include industry-specific urgency terms (e.g., "P1", "blocker", "compliance deadline") to the High category.</li><li><strong>Change the field</strong>: Use your existing priority or urgency field — just update the field name and values in the prompt.</li><li><strong>Adjust the threshold</strong>: If you want the agent to always set a priority (never ask), remove Logic Rule #4 and default ambiguous tasks to Normal.</li><li><strong>Expand to severity</strong>: Add a second field like "Severity" alongside Priority for teams that track both.</li></ul><h2 data-id="expected-results"><strong>Expected Results</strong></h2><ul><li><strong>Immediate:</strong> Every new task gets triaged within seconds of creation — no manual scanning required.</li><li><strong>Team impact:</strong> Managers stop spending time reading every description to assign priority; the backlog stays clean.</li><li><strong>Edge case safety:</strong> Ambiguous tasks don't get silently mislabeled — the agent asks instead of guessing.</li></ul><p>Give this agent a try in your intake or operations workflows and see how much manual triage you can eliminate 💯</p>]]>
        </description>
    </item>
    <item>
        <title>Content Repurposer Agent - Turn One Blog into Multi-Channel Assets</title>
        <link>https://community.wrike.com/discussion/2779/content-repurposer-agent-turn-one-blog-into-multi-channel-assets</link>
        <pubDate>Thu, 05 Mar 2026 12:35:43 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2779@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This Wrike AI agent turns any blog post in a task description into four channel-ready marketing assets (LinkedIn post, newsletter blurb, social snippet, and ad copy variations) with a single checkbox trigger, saving content marketers hours of manual repurposing while keeping a consistent, customizable brand voice across all outputs. </p><p></p><p>Hello Community! 👋</p><p>Today we’re bringing a guide that shows you how to set up a <strong>Content Repurposer Agent</strong> that reads a blog post from a task description and generates four ready-to-use marketing assets: a LinkedIn post, a newsletter blurb, a social snippet, and ad copy variations.</p><p>Below, you’ll find the complete configuration, prompts, setup instructions, and tips so you can adapt this to your own workflow ✔️</p><h2 data-id="agent-goal-use-case"><strong>Agent Goal / Use Case</strong></h2><p>Marketing teams write long-form blog posts but then spend hours manually adapting them for other channels such as LinkedIn, email newsletters, social media, and paid ads. Each format has different length limits, tone expectations, and structural requirements.</p><p>The <strong>Content Repurposer Agent</strong> reads the blog post, identifies the hook, key value proposition, and call to action, then generates all four assets in one go, posted as a comment on the task.</p><h2 data-id="how-it-works"><strong>How It Works</strong></h2><p>A single agent handles the workflow:</p><ul><li><strong>Trigger:</strong> When a checkbox custom field is checked (for example, “Generate Marketing Assets”).</li><li><strong>Action:</strong> Posts a comment on the task with four formatted marketing assets generated from the task description.</li></ul><p><strong>Time Saved:</strong> 2–4 hours/week per content marketer.</p><h2 data-id="prerequisites"><strong>Prerequisites</strong></h2><ul><li><strong>Custom Field:</strong> “Generate Marketing Assets” (or similar).</li><li><strong>Type:</strong> Checkbox</li><li><strong>Location:</strong> Add this field to the relevant content/blog project or folder.</li><li><strong>Content Source:</strong> Ensure the blog content is stored in the task Description field.</li></ul><h2 data-id="agent-content-repurposer"><strong>Agent: Content Repurposer</strong></h2><p>This agent analyzes the task description (the blog post) and generates four distinct marketing assets in a single comment.</p><h3 data-id="configuration"><strong>Configuration:</strong></h3><div><table><colgroup><col /><col /><col /></colgroup><tr><th scope="col"><p><strong>Setting</strong></p></th><th scope="col"><p><strong>Value</strong></p></th></tr><tr><td><p>Name</p></td><td><p>Content Repurposer – Blog to Multi-Channel</p></td></tr><tr><td><p>Scope</p></td><td><p>The work item where the agent was added</p></td></tr><tr><td><p>Trigger</p></td><td><p>Custom field changed – checkbox set to checked</p></td></tr><tr><td><p>Action</p></td><td><p>Post a comment</p></td></tr></table></div><h3 data-id="prompt"><strong>Prompt:</strong></h3><p><strong>General instructions:</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/3MDPCGO35D6K\/content-1-gimp.png&quot;,&quot;name&quot;:&quot;Content 1 GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:78029,&quot;width&quot;:800,&quot;height&quot;:527,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F3MDPCGO35D6K%2Fcontent-1-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3604,&quot;dateInserted&quot;:&quot;2026-03-05T12:29:34+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F3MDPCGO35D6K%2Fcontent-1-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png" alt="Content 1 GIMP.png" height="527" width="800" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/3MDPCGO35D6K/content-1-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<p></p><div><div><p>ROLE: Brand Voice Content Specialist</p><p></p><p>OBJECTIVE: Analyze the Task Description (the blog post) and generate four</p><p>distinct assets: a LinkedIn Post, a Newsletter Blurb, a Social Snippet,</p><p>and Ad Copy.</p><p></p><p>BRAND VOICE GUIDELINES:</p><p>- Tone: [Insert your tone here, e.g., "Professional yet witty",</p><p>  "Authoritative", or "Casual/Energetic"]</p><p>- Style: Use active voice, avoid industry jargon, use short punchy</p><p>  sentences</p><p>- Formatting: Use emojis sparingly but effectively for bullet points</p><p></p><p>INPUT DATA:</p><p>- The Task Description (primary source — this is the blog post)</p><p>- Custom Fields (optional: if a "Target Audience" field exists, tailor</p><p>  the tone accordingly)</p></div></div><p><br /><strong>Action: </strong></p><ul><li>Post a comment</li></ul><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/6CUCD0DCZNPK\/action-post-a-comment-gimp.png&quot;,&quot;name&quot;:&quot;Action - post a comment GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:85105,&quot;width&quot;:784,&quot;height&quot;:511,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F6CUCD0DCZNPK%2Faction-post-a-comment-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3605,&quot;dateInserted&quot;:&quot;2026-03-05T12:31:00+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F6CUCD0DCZNPK%2Faction-post-a-comment-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png" alt="Action - post a comment GIMP.png" height="511" width="784" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/6CUCD0DCZNPK/action-post-a-comment-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p><br />
CORE LOGIC:<br /><br />
1. Read the entire Task Description. Identify the "Hook," the</p><p>   "Key Value Proposition," and the "Call to Action" (CTA).<br /><br />
2. LinkedIn Post: Create a post (max 200 words) with a strong opening</p><p>   hook, 3 bulleted takeaways, and a CTA to read the full blog.<br /><br />
3. Newsletter Blurb: Write a 2–3 sentence teaser designed to drive</p><p>   clicks from an email campaign.<br /><br />
4. Social Snippet: Create a tweet-style quote or insight</p><p>   (under 280 characters).<br /><br />
5. Ad Copy: Create 3 variations of headlines and short body text</p><p>   (max 50 words each) for a paid social ad.<br /><br />
OUTPUT FORMAT:<br /><br />
### 📝 LinkedIn Post</p><p>[Generated LinkedIn post]<br /><br />
### 📧 Newsletter Blurb</p><p>[Generated newsletter teaser]<br /><br />
### 💬 Social Snippet</p><p>[Generated tweet-style snippet]<br /><br />
### 📢 Ad Copy</p><p>**Variation 1:** [Headline] — [Body]</p><p>**Variation 2:** [Headline] — [Body]</p><p>**Variation 3:** [Headline] — [Body]<br /><br />
CONSTRAINTS &amp; GUIDELINES:</p><p>- Do not include internal reasoning — output only the final marketing</p><p>  copy</p><p>- Each asset must be self-contained and ready to copy-paste</p><p>- Maintain consistent brand voice across all four assets</p><p>- If the Task Description contains fewer than 100 words, do not generate</p><p>  copy. Instead post: "This content appears too short for a full</p><p>  transformation. Please provide the complete blog post."</p></div></div><p></p><h2 data-id="deployment-steps"><strong>Deployment Steps</strong></h2><h3 data-id="step-1"><strong>Step 1</strong></h3><ul><li>Go to Space Settings → Custom Fields</li><li>Create a Checkbox field called "<strong>Generate Marketing Assets</strong>" (or reuse an existing trigger field)</li><li>Add it to your content/blog project</li></ul><h3 data-id="step-2"><strong>Step 2</strong></h3><ul><li>Go to Space Settings → AI Agents → + Custom AI agent</li><li>Configure name, scope, trigger, and action as above</li><li>Paste the prompt</li></ul><h3 data-id="step-3"><strong>Step 3</strong></h3><ul><li>Test in Playground with a sample blog post to verify all four assets generated correctly.</li></ul><h3 data-id="step-4"><strong>Step 4</strong></h3><ul><li>Appoint the agent to your content/blog folder or project.</li></ul><h2 data-id="expected-results"><strong>Expected Results</strong></h2><ul><li>Blog posts get four ready-to-use marketing assets in seconds.</li><li>Consistent brand voice across all channels without manual adaptation.</li><li>Content marketers spend time refining rather than drafting from scratch.</li><li>Easy to customize: swap out the Brand Voice Guidelines to match your organization's tone.</li></ul><h2 data-id="tips"><strong>Tips</strong></h2><ul><li><strong>Customize the tone</strong>: Replace the placeholder in BRAND VOICE GUIDELINES with your actual brand voice (e.g., "Friendly and expert, like explaining to a smart colleague").</li><li><strong>Add a Target Audience field</strong>: If you have a "<strong>Target Audience</strong>" custom field on your tasks, the agent will tailor its output accordingly.</li><li><strong>Iterate on the ad copy</strong>: The 3 variations give you A/B test options, pick the best one or mix and match.</li></ul><p>Give this agent a try and let us know what you think! 🤗</p>]]>
        </description>
    </item>
    <item>
        <title>✔️ AI Agents February Update: GA Release, Multi-Action Agents, and Database Lookups</title>
        <link>https://community.wrike.com/discussion/2700/%EF%B8%8F-ai-agents-february-update-ga-release-multi-action-agents-and-database-lookups</link>
        <pubDate>Tue, 17 Feb 2026 10:35:02 +0000</pubDate>
        <category>Wrike AI</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2700@/discussions</guid>
        <description><![CDATA[<p>TL;DR: AI agents are now generally available for all Business+ accounts, so there’s no need to access them in Wrike Labs anymore. The biggest new capability is multi-action agents, which let you combine multiple actions in a single agent without chaining. Agents can now also look up records in linked databases. Featured example: A “revision tracker” agent that counts and documents review cycles. It can be deployed in about five minutes and replaces what previously required a recipe in an integration platform.</p><p></p><p>Hey Community! 👋</p><p>Welcome to <strong>Wrike Agent News</strong>! <br /><br />
This monthly digest features updates, learnings, and how-tos from the Wrike R&amp;D team.</p><p>This month is a milestone: <strong>AI agents are now generally available</strong> 💯  Everything you’ve been testing in Labs is production-ready and available on Business+ plans. Thank you! And the feature you requested most, <a href="https://community.wrike.com/discussion/2628/releases-multi-action-ai-agents-updates-for-template-solution-center-apex-plan-01-26-2026?utm_source=community-search&amp;utm_medium=organic-search&amp;utm_term=multi+action+agent#:~:text=AI%20Agents%3A%20Multi%2DAction%20Agents%20Now%20Available" target="_blank" rel="nofollow noopener ugc"><strong>multi-action agents</strong></a>, is also here.</p><h3 data-id="generally-available">🚀 <strong>Generally available:</strong> </h3><p><strong>AI agents are out of Labs.</strong></p><p>Agents are now enabled by default on all Business+ accounts. Admins can manage the feature in <strong>Settings → Generative AI</strong>.</p><p>What this means:</p><p>• No more Labs opt-in required.</p><p>• Agents are production-ready and supported.</p><p>• Admin control via account settings toggle.</p><p>• Available on Business, Enterprise, and Pinnacle plans.</p><p>If you were already running agents in Labs, nothing changes; your existing agents keep working. If you haven’t started yet, now’s a great time.</p><h3 data-id="multi-action-agents">🎯 <strong>Multi-action agents</strong></h3><p><strong>One agent. Multiple actions. No more chaining for independent steps.</strong></p><p>Agents can now execute two or more actions from a single trigger. Add a prefix, an assignment, and a comment all inside one agent.</p><p><strong>Use case:</strong> New task created → agent classifies the request, renames it with a prefix like “[DESIGN]”, assigns the right specialist, and posts a summary comment. One agent replaces what previously required three chained agents.</p><p><strong>Use case:</strong> Status changes to “Signed Off” → agent assigns the next owner and updates a custom field with the sign-off date. Two actions, zero handoffs.</p><p>Each action gets its own prompt, plus a new <strong>General Instruction</strong> field that is prepended to every action. Use the general instruction for shared context: what the agent’s role is, what it should look at, and reference data like team rosters or category lists. Then, each action prompt focuses on its specific job: classify, rename, or assign. Write the context once, and every action has it.</p><p><em><strong>Note:</strong></em><em> Multi-action means independent actions running in parallel. If Action B needs the result of Action A, you still need agent chaining. Multi-action is for actions that don’t depend on each other.</em></p><h3 data-id="link-to-database-action">🔗 <strong>Link-to-database action</strong></h3><p><strong>Agents can now look up and select database records.</strong></p><p>Two database upgrades:</p><ul><li><strong>Read:</strong> Agents can see <strong>Link to Database </strong>fields and <strong>Mirror</strong> fields when analyzing a task, including data pulled from linked records, not just the task’s own fields.<br /><br /></li><li><strong>Write:</strong> When a task mentions a client name, project code, or equipment ID, the agent can search your linked database and select the matching record. This automatically populates all associated mirror fields.</li></ul><p><strong>Use case:</strong> New support ticket mentions “Acme Corp” → agent finds the matching record in your client database, selects it, and mirror fields auto-fill with account manager, SLA tier, and contract details.</p><p><strong>Use case:</strong> Task description contains equipment ID → agent matches it in the equipment database and populates warranty status, location, and vendor info.</p><h3 data-id="under-the-hood">🛠️ <strong>Under the hood</strong></h3><p><strong>Smarter context, better reliability.</strong></p><ul><li><strong>Improved retrieval:</strong> Agents now receive cleaner, more relevant context. Irrelevant data is filtered out before the agent reasons, leading to more accurate results.</li><li><strong>Empty fields visible:</strong> Agents can now see empty custom fields, so they can correctly identify what’s missing and needs filling.</li><li><strong>Language respect:</strong> Agents respond in the same language as their instructions, write your prompt in German, get results in German.</li><li>Peak load handling: Backend improvements ensure agents perform reliably even during high-volume periods.</li></ul><h3 data-id="agent-of-the-month-revision-tracker">Agent of the month: Revision tracker</h3><p><strong>In this section, we’ll dive into a specific use case.</strong></p><p><strong>Use case:</strong> Teams that collaborate with clients or external stakeholders, agencies, professional services, legal review teams often send work back and forth multiple times before it’s finalized. Tracking how many revision cycles each item goes through is valuable for identifying bottlenecks, managing client expectations, and scoping future work.</p><p>“We needed to know how many times a campaign was returned to the client for more information. Before, this would have required a recipe in our integration platform. Now, it’s a three-sentence agent.”</p><p><strong>The solution:</strong> A single multi-action agent with two actions:</p><ol><li><strong>Action 1 (Update field): </strong>Increments a numeric custom field (“Revision Count”) each time the status returns to a review stage. If the field is empty, it sets it to 1. If it already has a value, it adds 1.<br /></li><li><strong>Action 2 (Comment): </strong>Posts a note recording the revision, including which cycle this is and a brief summary of what the agent observed in the task at the time of the status change.</li></ol><p><strong>Time saved:</strong> Eliminates manual tracking and replaces what previously required a recipe in an integration platform.<br /><strong>Prerequisites:</strong> A numeric custom field (e.g., “Revision Count”), a workflow with a client-facing review status.<br /><strong>Deployment time:</strong> 5 minutes: the prompt is just a few sentences.</p><p><strong>How it works:</strong></p><ol><li><strong>Trigger fires:</strong> Status changes to the review stage (e.g., “Client Update Needed”).</li><li><strong>Counts:</strong> Agent reads the current revision count and increments it by one.</li><li><strong>Documents:</strong> Agent posts a comment noting the revision cycle number and context.</li></ol><p>Two actions. Automatic revision history. No integration platform required.</p><p>Click the link below to view the solution and step-by-step configuration.</p><p>👉 <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F1My0elyO_QcZuEIxg_f0ROx3M8uJ6Z6xy8st4p99PkpY%2Fedit%3Fusp%3Dsharing" target="_blank" rel="nofollow noopener ugc"><strong>Solution and configuration</strong></a></p><p><strong>The complete guide includes:</strong></p><ul><li>General instructions and action prompts with customization instructions.</li><li>Step-by-step deployment walkthrough.</li><li>Examples for agency/client review, legal redlining, and creative approval workflows.</li><li>Tips for adapting the counter pattern to other use cases (e.g., escalation counting, reassignment tracking).</li></ul><h3 data-id="combining-this-month-s-features"><strong>Combining this month’s features</strong></h3><p>Multi-action agents and agent chaining complement each other. Use multi-action for independent steps, chaining for dependent ones.</p><p><strong>Complete task lifecycle:</strong> Status changes to review stage → <strong>Revision tracker</strong> counts the cycle and documents context. If a task gets stuck, the <strong>stale task escalation</strong> chain from December catches it.</p><p>Combine a multi-action intake agent (to triage new work) with the revision tracker (to monitor ongoing cycles) and the escalation chain (to catch stalled items), three agents covering the full lifecycle.</p><h3 data-id="questions"><strong>Questions?</strong></h3><p><strong>Want help setting up the revision tracker?</strong> Reach out to us in the comments.</p><p><strong>Need a different agent type?</strong> Contact your Customer Success Manager.</p><p><strong>Found a bug or have feedback?</strong> Use the feedback link in the product or let us know in the comments.</p><h3 data-id="on-our-radar"><strong>On our radar</strong></h3><p>Areas we’re actively exploring:</p><ul><li><strong>Sequential action dependencies:</strong> Actions where Step B uses the result of Step A.</li><li><strong>More trigger types:</strong> Expanding what can kick off an agent.</li><li><strong>Expanded field support:</strong> Broader coverage of field types that agents can read and write.</li></ul><p>Your feedback shapes our roadmap. Keep testing, keep sharing what works (and what doesn’t).</p>]]>
        </description>
    </item>
    <item>
        <title>AI Agents Index: Copy‑Paste Prompts, Configs &amp; Implementation Guides</title>
        <link>https://community.wrike.com/discussion/2762/ai-agents-index-copy-paste-prompts-configs-implementation-guides</link>
        <pubDate>Tue, 03 Mar 2026 11:48:40 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2762@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This post is a central index of Wrike AI agents shared by the community and Wrike team, each with copy‑paste prompts, configuration details, and implementation guides so you can quickly discover, set up, and reuse AI automations tailored to different workflows.</p><p><br />
Hello Community! 👋</p><p><br />
This post is a central index for all AI agents shared by the Community and the Wrike team. Each linked post includes the full prompt, configuration details, and usage notes.</p><h2 data-id="how-to-use-this-index">How to Use This Index</h2><ul><li>Browse the list below to find an agent that fits your use case. <br />
 </li><li>Click the link to open the main post for the full prompt, configuration, and implementation details.</li></ul><h2 data-id="index">Index:</h2><p><strong>1. Revision Tracker Agent – </strong>Multi-action agent that automatically counts how many times a task is sent back for review by incrementing a numeric “Revision Count” custom field and posting a short comment whenever a task enters a selected review status. Provides dashboard-ready revision data and replaces manual tracking, with customizable field, trigger status, and comment format.<strong>  </strong></p><p><br /><strong> Link: </strong><a href="https://community.wrike.com/discussion/2713/revision-tracker-agent#latest" target="_blank" rel="nofollow noopener ugc"><strong>Revision Tracker Agent</strong></a></p><p><br /><strong>2.Briefing Agent – Target Audience Profiler – </strong>Reads the request description and automatically fills in the “Target Audience” multi-select custom field with the right segments (e.g., SMB, Enterprise, Marketing Leaders, IT Admins). Removes the need for managers to manually interpret and tag the audience, saving time and keeping targeting consistent whenever a new request is created or its description is updated. <strong> </strong></p><p><strong>Link:  </strong><a href="https://community.wrike.com/discussion/2719/briefing-agent-target-audience-profiler#latest" target="_blank" rel="nofollow noopener ugc"><strong>Briefing Agent – Target Audience Profiler</strong></a></p><p><br /><strong>3.Stale Task Escalation Agent –</strong> Monitors tasks that sit too long without progress and automatically escalates them. When a task remains in a selected status beyond your defined time threshold, the agent can @mention owners or managers, post a reminder comment, or update fields/status for visibility. This helps prevent work from getting stuck and keeps your pipeline moving without manual follow-up. </p><p> <br /><strong>Link: </strong><a href="https://community.wrike.com/discussion/2736/stale-task-escalation-agent#latest" target="_blank" rel="nofollow noopener ugc"><strong>Stale Task Escalation Agent</strong></a></p><p></p><p><strong>4.Content Repurposer Agent – Turn One Blog into Multi-Channel Assets</strong> - Reads the blog post in the task description and automatically generates four channel-ready marketing assets (LinkedIn post, newsletter blurb, social snippet, and ad copy variations) when a checkbox custom field is selected. Eliminates the need for manual repurposing, saving content marketers hours each week while maintaining a consistent, customizable brand voice across all outputs.</p><p><strong>Link:</strong> <a href="https://community.wrike.com/discussion/2779/content-repurposer-agent-turn-one-blog-into-multi-channel-assets" target="_blank" rel="nofollow noopener ugc"><strong>Content Repurposer Agent - Turn One Blog into Multi-Channel Assets</strong></a></p><p></p><p><strong>5.Urgency Classifier Agent</strong> – Automatically scans new tasks for urgency cues in the title and description, then updates the Priority custom field to High, Normal, or Low. When the request is unclear or conflicting, it posts a comment asking the assignee to clarify instead of guessing. Helps teams reduce manual triage, speed up intake, and avoid mislabeling ambiguous requests.</p><p>Link: <a href="https://community.wrike.com/discussion/2822/urgency-classifier-agent#latest" target="_blank" rel="nofollow noopener ugc"><strong>Urgency Classifier Agent</strong></a></p>]]>
        </description>
    </item>
    <item>
        <title>AI Agents Library &amp; Inspiration Space Is Now Live!</title>
        <link>https://community.wrike.com/discussion/2756/ai-agents-library-inspiration-space-is-now-live</link>
        <pubDate>Mon, 02 Mar 2026 12:51:53 +0000</pubDate>
        <category>Announcements</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2756@/discussions</guid>
        <description><![CDATA[<p>TL;DR: We’ve launched an AI Agents Library &amp; Inspiration space where you can explore, copy, and customize Wrike AI Agent configurations (prompts, triggers, actions), see real example agents like Revision Tracking and Stale Task Escalation, and share your own setups and use cases with the Community.</p><p><br />
Hello Community! 👋</p><p><br />
We’ve launched a new dedicated Community space: <a href="https://community.wrike.com/categories/ai-agents-library-inspiration" target="_blank" rel="nofollow noopener ugc"><strong>AI Agents Library &amp; Inspiration</strong></a> 🎉<br />
This is where you can explore Wrike Team and Community shared AI Agent prompts, configurations, and implementation tips.<br /></p><h2 data-id="what-you-ll-find-there">What you’ll find there</h2><p><br />
In the AI Agents Library &amp; Inspiration space, you can explore:<br /></p><ul><li>Full AI Agent configurations from the Wrike team &amp; Community.  </li><li>Copy‑ready prompts for different use cases. </li><li>Trigger and action setups so you can recreate agents in your own account.  </li><li>Real examples and implementation guides you can reuse, adapt, and improve.  </li></ul><p><br />
We’re continuously adding new agents, including things like:<br /></p><ul><li>Revision Tracking Agent</li><li>Briefing Agent</li><li>Stale Task Escalation Agent</li></ul><p><br />
…and more to come.<br /></p><h2 data-id="how-you-can-participate">How you can participate</h2><p><br />
We’d love for you to use this space not just as a library, but as a place to learn from each other:<br /></p><ul><li>Copy and customize the agents shared by the Wrike team and Community members.  </li><li>Post your own AI Agents (prompt + triggers + actions).  </li><li>Describe your use cases and the problems your agents solve.  </li><li>Ask questions and share feedback on configurations or results.  </li></ul><p><br /><strong>👉 </strong><a href="https://community.wrike.com/categories/ai-agents-library-inspiration" target="_blank" rel="nofollow noopener ugc"><strong>Visit the AI Agents Library &amp; Inspiration space</strong></a><br />
Looking forward to seeing how you put Wrike AI Agents to work in your workflows 🙌</p>]]>
        </description>
    </item>
    <item>
        <title>Revision Tracker Agent</title>
        <link>https://community.wrike.com/discussion/2713/revision-tracker-agent</link>
        <pubDate>Fri, 20 Feb 2026 06:06:48 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2713@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This multi-action Wrike AI agent automatically tracks how many times a task is sent back for review by incrementing a numeric “Revision Count” custom field and posting a short comment each time a task enters a selected review status. It gives you dashboard-ready data on revision cycles, replaces manual tracking or external integrations, and is fully customizable to your workflow (field name, trigger status, and comment format).</p><p>Hello Community! 👋</p><p>Today, we’re excited to share a <strong>Revision Tracker Agent (multi-action) </strong>that automatically tracks your review cycles directly in Wrike without any extra tools needed. Below, you’ll find the complete configuration, the prompts to use, setup instructions, and customization tips so you can use it in your own workflow ✔️</p><h2 data-id="agent-goal-use-case"><strong>Agent Goal / Use Case</strong></h2><p>Automatically count how many times a task returns to a <strong>review status</strong> and log each revision cycle with a comment, all from a single trigger and without using an external integration platform.</p><p><strong>Typical scenario:</strong></p><p>Teams that collaborate with clients or external stakeholders (agencies, professional services, legal review, creative production, etc.) send work back and forth multiple times before it’s finalized.</p><p>A campaign goes to the client for review, comes back with feedback, gets revised, goes back out and so on. Knowing how many revision cycles each item goes through helps you:</p><ul><li>Spot bottlenecks</li><li>Manage client expectations</li><li>Improve scoping and estimation for future work</li></ul><p><strong>Time Saved: </strong></p><ul><li>Eliminates manual revision tracking</li><li>Replaces what previously required a recipe in an integration platform</li></ul><p><strong>Prerequisites:</strong></p><p>Before setting up this agent, you’ll need:</p><ul><li>A numeric custom field (e.g., Revision Count)</li><li>A workflow with a client- or stakeholder-facing review status (e.g., “Client Review,” “Client Update Needed”)</li></ul><h2 data-id="how-it-works"><strong>How It Works</strong></h2><p>This is a <a href="https://community.wrike.com/discussion/2628/releases-multi-action-ai-agents-updates-for-template-solution-center-apex-plan-01-26-2026#latest:~:text=Multi%2DAction%20Agents%20Now%20Available" target="_blank" rel="nofollow noopener ugc"><strong>multi-action agent</strong></a> that runs off a <strong>single trigger</strong> and performs <strong>two actions</strong> in parallel:</p><ul><li><strong>Reads the trigger</strong></li><li>The agent runs when a task enters your chosen review status (for example, “Client Update Needed”).</li><li><strong>Increments the counter</strong></li><li>The agent reads the current value in the Revision Count custom field.</li><li>If the field has a value, it adds +1.</li><li>If it’s empty, it sets the value to 1.</li><li><strong>Documents the cycle</strong></li><li>The agent posts a comment on the task noting the current revision cycle (e.g., “Revision #3: Task has returned to client review status.”).</li></ul><p>Because the field update and the comment don’t depend on each other, they can run in parallel as a multi-action agent.</p><p><strong>Prompt (Complete Configuration)</strong>This agent uses a General Instruction (shared context for all actions) plus two separate Action Prompts.</p><h2 data-id="step-1-copy-these-prompts"><strong>Step 1: Copy these Prompts 👇</strong> </h2><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/2SV0MKJ0G8IV\/final-gi-gimp.png&quot;,&quot;name&quot;:&quot;Final GI  GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:63721,&quot;width&quot;:804,&quot;height&quot;:361,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F2SV0MKJ0G8IV%2Ffinal-gi-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3553,&quot;dateInserted&quot;:&quot;2026-02-20T06:05:33+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2F2SV0MKJ0G8IV%2Ffinal-gi-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png" alt="Final GI  GIMP.png" height="361" width="804" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/2SV0MKJ0G8IV/final-gi-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>You are a Revision Tracking Agent.<br /><br />
Your job is to track how many times a task returns to a review status</p><p>by maintaining a revision counter and documenting each cycle.<br /><br />
=== REVISION TRACKING ===<br /><br />
Counter field: "Revision Count" (numeric custom field)<br /><br />
When this agent is triggered, it means the task has entered a review</p><p>cycle. Read the task's title, description, and current custom field</p><p>values to understand what the task is about and what state it's in.</p></div></div><p><strong>Action 1: Change Custom Field</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/PHLYCUPJHT66\/action-1-final-gimp.png&quot;,&quot;name&quot;:&quot;Action 1 FINAL Gimp.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:59338,&quot;width&quot;:749,&quot;height&quot;:509,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FPHLYCUPJHT66%2Faction-1-final-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3554,&quot;dateInserted&quot;:&quot;2026-02-20T06:06:00+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FPHLYCUPJHT66%2Faction-1-final-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png" alt="Action 1 FINAL Gimp.png" height="509" width="749" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/PHLYCUPJHT66/action-1-final-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>Look at the current value of the "Revision Count" field.<br /><br />
If the field is empty, set it to 1.</p><p>If the field already has a value, add 1 to the current value.</p></div></div><p><strong>Action 2: Post a Comment</strong></p><span data-embedjson="{&quot;url&quot;:&quot;https:\/\/us.v-cdn.net\/6038595\/uploads\/IYT5S9EKGM6X\/action-2-final-gimp.png&quot;,&quot;name&quot;:&quot;Action 2 FINAL GIMP.png&quot;,&quot;type&quot;:&quot;image\/png&quot;,&quot;size&quot;:88909,&quot;width&quot;:744,&quot;height&quot;:690,&quot;displaySize&quot;:&quot;large&quot;,&quot;float&quot;:&quot;none&quot;,&quot;downloadUrl&quot;:&quot;https:\/\/community.wrike.com\/api\/v2\/media\/download-by-url?url=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FIYT5S9EKGM6X%2Faction-2-final-gimp.png&quot;,&quot;active&quot;:true,&quot;mediaID&quot;:3555,&quot;dateInserted&quot;:&quot;2026-02-20T06:06:16+00:00&quot;,&quot;insertUserID&quot;:5783,&quot;foreignType&quot;:&quot;embed&quot;,&quot;foreignID&quot;:&quot;5783&quot;,&quot;embedType&quot;:&quot;image&quot;,&quot;embedStyle&quot;:&quot;rich_embed_card&quot;}">
    <span>
        <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fus.v-cdn.net%2F6038595%2Fuploads%2FIYT5S9EKGM6X%2Faction-2-final-gimp.png" rel="nofollow noopener ugc" target="_blank">
            <img src="https://us.v-cdn.net/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png" alt="Action 2 FINAL GIMP.png" height="690" width="744" data-display-size="large" data-float="none" data-type="image/png" data-embed-type="image" srcset="https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=300, width=300/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png 300w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=600, width=600/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png 600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=800, width=800/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png 800w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1200, width=1200/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png 1200w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=1600, width=1600/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png 1600w, https://us.v-cdn.net/cdn-cgi/image/quality=80, format=auto, fit=scale-down, height=2000, width=2000/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png 2000w, https://us.v-cdn.net/6038595/uploads/IYT5S9EKGM6X/action-2-final-gimp.png" sizes="100vw" /></a>
    </span>
</span>
<div><div><p>Post a brief comment documenting this revision cycle.<br /><br />
FORMAT:<br /><br />
"Revision [N] — [one-line summary of the task's current state]"<br /><br />
Where [N] is the current revision cycle number (the value you would</p><p>set the Revision Count field to).<br /><br />
Keep it short. One or two sentences describing what the task is about</p><p>and what information is currently present. Do not add action items or</p><p>instructions — just document the state of the task at this point in</p><p>the cycle.</p></div></div><h2 data-id="step-2-customize-for-your-workflow"><strong>Step 2: Customize for Your Workflow</strong></h2><p>Update these sections in the prompts:</p><p><strong>Revision Count field name (Required)</strong></p><p>Create a numeric custom field called "Revision Count" (or whatever name fits your workflow). Make sure it's available in the folder or project where the agent will run. Update the field name in the General Instruction if you use a different name.</p><p><strong>Trigger status (Required)</strong></p><p>Identify the status in your workflow that represents "sent back for review." Common examples:</p><ul><li>"Client Update Needed"</li><li>"Returned for Revision"</li><li>"Changes Requested"</li><li>"Back to Client"</li></ul><p>This is the status that triggers the agent. Every time a task enters this status, the counter increments.</p><p><strong>Comment format (Optional)</strong>Adjust the Action 2 prompt to include additional context relevant to your team:</p><ul><li>Add a mention of the assignee: "Revision [N] — assigned to [name] — [summary]"</li><li>Tag a stakeholder for visibility.</li><li>Include the previous status the task came from.</li></ul><h2 data-id="step-3-deploy-to-wrike"><strong>Step 3: Deploy to Wrike</strong></h2><ol><li><strong>Navigate to your space settings</strong> → AI agents tab</li><li><strong>Click "+ Custom AI agent"</strong></li><li><strong>Configure:</strong><ul><li><strong>Name:</strong> Revision Tracker</li><li><strong>Scope:</strong> "Any subitem of the item where the agent was added"</li><li><strong>Trigger:</strong> "Status changed" (to your review status, e.g., "Client Update Needed")</li><li><strong>General Instruction:</strong> Paste the General Instruction from Step 1</li><li><strong>Action 1:</strong> Update custom field → select "Revision Count" → paste the Action 1 prompt</li><li><strong>Action 2:</strong> Post comment → paste the Action 2 prompt</li></ul></li><li><strong>Test in Playground:</strong><ul><li>Select a task that has been through at least one review cycle.</li><li>Verify the agent correctly reads the current Revision Count value.</li><li>Confirm it increments by 1 (or sets to 1 if empty).</li><li>Check that the comment is concise and accurately describes the task state.</li><li>Test with both an empty Revision Count field and an existing value.</li></ul></li><li><strong>Deploy:</strong><ul><li>Appoint the agent to your project or triage folder.</li><li>The agent activates every time any task in that folder enters the trigger status.</li><li>Monitor the Agent Activity Dashboard for the first few triggers to confirm the count increments correctly.</li></ul></li></ol><h2 data-id="why-two-actions-instead-of-one"><strong>Why Two Actions Instead of One?</strong></h2><p>You might wonder why this isn't a single "post comment" agent that also mentions the revision number. Splitting into two actions gives you:</p><ul><li><strong>Dashboard-ready data:</strong> A numeric field means you can add "Revision Count" as a column in any table view, sort by it, filter by it, and build charts. A comment alone can't do that.</li><li><strong>Reporting at scale:</strong> "Which items had more than 3 revision cycles this quarter?" is a one-click filter with a custom field. With comments, you'd need to read every thread.</li><li><strong>Audit trail:</strong> The comment provides human-readable context (what happened during this cycle), while the field provides machine-readable data (how many cycles total). Both together tell the full story.</li><li><strong>Composability:</strong> Other agents or automations can read the Revision Count field to make decisions (e.g., "if Revision Count &gt; 5, escalate to manager").</li></ul><h2 data-id="expected-results"><strong>Expected Results</strong></h2><p><strong>Immediate Impact:</strong></p><ul><li>Every revision cycle is automatically counted and documented.</li><li>No manual tracking, the agent handles it on every status change.</li><li>Historical record of revision patterns across all tasks.</li></ul><p><strong>Team Benefits:</strong></p><ul><li>Visibility into which items are churning (sort by Revision Count).</li><li>Data for retrospectives : "why did this campaign go through 7 revision cycles?".</li><li>Client conversations backed by data : "this project had 5 rounds vs. the typical 2".</li><li>Pair with dashboards: add Revision Count as a column and instantly see which items cycle most: a leading indicator of scope creep or unclear requirements.</li></ul><h2 data-id="troubleshooting"><strong>Troubleshooting</strong></h2><p><strong>Counter doesn't increment:</strong></p><ul><li>Verify the trigger is set to the correct status (exact match).</li><li>Check that the agent scope includes the items you expect (subitems vs. parent items).</li><li>Confirm the "Revision Count" custom field is accessible in the folder where the agent runs.</li></ul><p><strong>Counter resets to 1:</strong></p><ul><li><strong>Check the Action 1 prompt:</strong>  It should read the current value before updating. If the prompt says "set to 1" without checking the existing value, the counter will always reset.</li></ul><p><strong>Comment is too verbose:</strong></p><ul><li><strong>Tighten the Action 2 prompt:</strong> "Keep it to one sentence" or "Maximum 50 words"</li></ul><p>Remove any language that asks the agent to analyze or recommend  the comment should only document.</p><p><strong>Agent fires on the wrong status:</strong></p><ul><li>Double-check the trigger configuration. The trigger should fire specifically when the status changes to your review status, not on any status change.</li></ul><p><strong>Counter shows on tasks that were never reviewed:</strong></p><ul><li>The field will only have a value if the agent has triggered at least once. If you see unexpected values, check whether someone manually set the field or if the agent scope is broader than intended..</li></ul>]]>
        </description>
    </item>
    <item>
        <title>Automating Request-forms</title>
        <link>https://community.wrike.com/discussion/2718/automating-request-forms</link>
        <pubDate>Mon, 23 Feb 2026 08:16:08 +0000</pubDate>
        <category>Product Feedback</category>
        <dc:creator>Nanna Larsen</dc:creator>
        <guid isPermaLink="false">2718@/discussions</guid>
        <description><![CDATA[<p>Hi,</p><p>I've looked through all of the automation actions, and I have an idea of another one, that would be very usefull. </p><p>If possible, when a trigger occours, fx the status of a task or project changes, then a pop up window with a request form could appear, allowing the user to be "forced" to answer it and thereby add information to the project.</p><p>The reason for this idea, is that in my team, we have set up the request forms in wrike with some external automations in our own system, and would like those to work together with wrikes automation, to create a more continuous flow with a limit to the manual steps.</p><p>It seems the automation in wrike is already spanding wide, so I hope it can include the request forms as an action as well.</p>]]>
        </description>
    </item>
    <item>
        <title>Briefing Agent – Target Audience Profiler</title>
        <link>https://community.wrike.com/discussion/2719/briefing-agent-target-audience-profiler</link>
        <pubDate>Mon, 23 Feb 2026 09:37:14 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Marco Bottaro</dc:creator>
        <guid isPermaLink="false">2719@/discussions</guid>
        <description><![CDATA[<h2 data-id="briefing-agent-target-audience-profiler"><strong>Briefing Agent – Target Audience Profiler</strong></h2><h3 data-id="overview"><strong>Overview</strong></h3><p>This guide shows you how to set up a Briefing Agent that analyzes request descriptions and auto-fills a “Target Audience” multi-select custom field based on predefined options.</p><p>Time Saved: 1–2 hours/week per marketing manager<br />
Prerequisites: Multi-select custom field for target audience<br />
Deployment Time: 15–20 minutes (one agent)</p><h3 data-id="the-use-case"><strong>The Use Case</strong></h3><p>Marketing request forms often include vague or incomplete audience details. Managers must read long descriptions and then manually tag target segments, which is inconsistent and time-consuming.<br />
The Briefing Agent reads the description and selects the right audience segments automatically.</p><h3 data-id="how-it-works"><strong>How It Works</strong></h3><p>A single agent:</p><ul><li>Trigger: When a new request is created or the description is updated</li><li>Action: Sets values in the “Target Audience” multi-select custom field based on the description</li></ul><h3 data-id="prerequisites"><strong>Prerequisites</strong></h3><ol><li>Custom Field: “Target Audience”<ul><li>Type: Multi-select</li><li>Values: Predefined audience options (e.g., “SMB,” “Enterprise,” “Marketing Leaders,” “IT Admins,” etc.)</li><li>Add to the relevant request/work intake workflow or space</li></ul></li></ol><h3 data-id="agent-briefing-agent-target-audience"><strong>Agent: Briefing Agent (Target Audience)</strong></h3><p>This agent analyzes the Item Description and selects one or more audience values from the existing field options.</p><h4 data-id="configuration">Configuration</h4><div><table><colgroup><col /><col /><col /></colgroup><tr><th><p>Setting</p></th><th><p>Value</p></th></tr><tr><td><p>Name</p></td><td><p>Briefing Agent – Target Audience Profiler</p></td></tr><tr><td><p>Scope</p></td><td><p>Any subitem of the item where the agent was added (e.g., campaign request folder/project)</p></td></tr><tr><td><p>Trigger</p></td><td><p>Field changed – Description (or on Item created + Description changed)</p></td></tr><tr><td><p>Action</p></td><td><p>Change custom field – “Target Audience” (multi-select)</p></td></tr></table></div><h4 data-id="prompt">Prompt</h4><p><code spellcheck="false" tabindex="0">ROLE: Strategic Audience Profiler  </code></p><p><code spellcheck="false" tabindex="0">OBJECTIVE: Populate the "Target Audience" multi-select field by analyzing the Item Description and selecting all relevant matches from the predefined field options.</code></p><p><code spellcheck="false" tabindex="0">OUTPUT FIELD/INTENT: Target Audience  </code></p><p><code spellcheck="false" tabindex="0">INPUT DATA:</code></p><p><code spellcheck="false" tabindex="0">- The Item Description (or Content Body)</code></p><p><code spellcheck="false" tabindex="0">- The list of available options defined in the "Target Audience" multi-select field  </code></p><p><code spellcheck="false" tabindex="0">CONSTRAINTS &amp; GUIDELINES:</code></p><p><code spellcheck="false" tabindex="0">- The output must consist ONLY of the selected values from the available options.</code></p><p><code spellcheck="false" tabindex="0">- Do not generate new audience types; you must strictly choose from the provided multi-select list.</code></p><p><code spellcheck="false" tabindex="0">- Select multiple options if the description supports multiple unique audience segments.</code></p><p><code spellcheck="false" tabindex="0">- Maintain the exact spelling and casing of the options as they appear in the field definition.</code></p><p><code spellcheck="false" tabindex="0">- Output must be a comma-separated list of the selected values.</code></p><p><code spellcheck="false" tabindex="0">- No introductory text, reasoning, or markdown formatting is allowed in the output.</code></p><p><code spellcheck="false" tabindex="0">CORE LOGIC:</code></p><p><code spellcheck="false" tabindex="0">1. Scan the "Target Audience" field metadata to identify all possible valid choices (the constrained list).</code></p><p><code spellcheck="false" tabindex="0">2. Read the Item Description thoroughly to understand the nature, utility, and intended user base of the item.</code></p><p><code spellcheck="false" tabindex="0">3. Evaluate each valid option from the list against the description.</code></p><p><code spellcheck="false" tabindex="0">4. Use best judgment to determine which options are explicitly or implicitly supported by the text.</code></p><p><code spellcheck="false" tabindex="0">5. Compile a list of all options that fit.</code></p><p><code spellcheck="false" tabindex="0">6. Output the final selection as a comma-separated string.</code></p><p><code spellcheck="false" tabindex="0">OUTPUT FORMAT:</code></p><p><code spellcheck="false" tabindex="0">The final output must be a clean, comma-separated list of the selected options.</code></p><h3 data-id="deployment-steps"><strong>Deployment Steps</strong></h3><ol><li>Go to Space Settings → Custom Fields<ul><li>Create/confirm multi-select field “Target Audience” and add audience values.</li></ul></li><li>Go to Space Settings → AI Agents → + Custom AI agent<ul><li>Configure name, scope, trigger, and action as above.</li><li>Paste the prompt.</li></ul></li><li>Test in Playground with sample requests to ensure the agent picks the right audiences.</li><li>Appoint the agent to your intake folder/project.</li></ol><h3 data-id="expected-results"><strong>Expected Results</strong></h3><ul><li>Requests arrive with audience segments already populated.</li><li>Less manual triage; more consistent segmentation for reporting and routing.</li></ul>]]>
        </description>
    </item>
    <item>
        <title>Welcome to the AI Agents Library &amp; Inspiration  🙌</title>
        <link>https://community.wrike.com/discussion/2711/welcome-to-the-ai-agents-library-inspiration</link>
        <pubDate>Fri, 20 Feb 2026 05:11:38 +0000</pubDate>
        <category>AI Agents Library &amp; Inspiration</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2711@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This space is for sharing real, working Wrike AI Agents you use in production. Post your agent’s goal/use case, prompt, triggers, actions, and any tips so others can reuse or adapt them. Browse configurations from the Wrike team and the community for inspiration. <br /><br /><br />
Hello Community! 😁<br /><br />
Welcome to the <strong>AI Agents Library &amp; Inspiration!</strong> 🎉  <br /><br />
This is a dedicated space to share Wrike AI Agent prompts, configurations, and how you’re using them to power your work.<br /><br /><strong>Here, you’ll find:</strong></p><ul><li>Full AI Agent configurations from the Wrike team  </li><li>Prompts used for different use cases  </li><li>Trigger setups and action scopes  </li><li>Real examples you can reuse, adapt, and improve</li></ul><p><br /><strong>We’d also love to see how you’re building and using AI Agents. Please feel free to:</strong></p><ul><li>Share your own AI Agent configurations (prompt + triggers + actions)  </li><li>Describe the use case and the problem your agent solves   </li><li>Get inspired by what other Community members are doing</li></ul><p><br /><strong>When sharing your AI Agents, it’s helpful to include:</strong></p><ul><li>Agent goal / use case  </li><li>Prompt (or main instructions)  </li><li>Trigger (how and when it runs)  </li><li>Actions (what it does in Wrike)  </li><li>Tips / lessons learned (optional but encouraged!)</li></ul><p><br />
👉 <strong>Important:</strong> Posts in this forum should be <strong>working AI Agents that you currently use in production</strong> and want to share with the Community.<br /><br />
If you have questions about AI Agents, or need help designing one, please use the main <a href="https://community.wrike.com/categories/34485002868759-Wrike-AI" target="_blank" rel="nofollow noopener ugc"><strong>Wrike AI forum</strong></a><strong> </strong>instead.<br /><br />
We’re excited to see what you create and how AI Agents help you streamline and scale your workflows. Let’s build smarter agents together! 🚀</p>]]>
        </description>
    </item>
    <item>
        <title>✍️ AI Agent Prompt Writing Guide</title>
        <link>https://community.wrike.com/discussion/2704/%EF%B8%8F-ai-agent-prompt-writing-guide</link>
        <pubDate>Wed, 18 Feb 2026 12:51:24 +0000</pubDate>
        <category>Wrike AI</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2704@/discussions</guid>
        <description><![CDATA[<p>TL;DR: This post shows how to write effective prompts for Wrike AI Agents so they behave like reliable team members. It explains what agents can and can’t access in Wrike, core prompt patterns (clear instructions, mapping tables, conditional rules, smart assignment, name extraction), and common pitfalls to avoid (vague asks, assumed context, chained actions, no fallbacks, math). It also includes testing tips for the Playground, ready-made prompt templates for typical workflows, and a short checklist to validate your prompt before going live.</p><p>Hi Community 👋</p><p>We’ve put together an <strong>AI Agent prompt writing guide</strong> to help you get more reliable, predictable results from your agents in Wrike. This post walks through how agents “think,” how to structure effective prompts, common pitfalls to avoid, and ready-to-use templates for real-world use cases like intake validation, smart assignment, and risk reporting.</p><h2 data-id="the-golden-rule"><strong>The Golden Rule </strong>💫</h2><p>Write prompts like you’re explaining a task to a new team member who is smart, but doesn’t know anything about your company yet. Be clear and specific about what you want them to do, but don’t over-prescribe how they should do it. Let the agent figure out the best way to get there based on the context it can see in Wrike.</p><h2 data-id="how-agents-think"><strong>How Agents Think</strong> 🤔</h2><p>Before writing prompts, understand what your agent can see:</p><h3 data-id="agents-can-read">Agents CAN read:</h3><ul><li>Task name, description, and comments.</li><li>Custom field values (text, numbers, drop-downs).</li><li>Current status and assignee.</li><li>User profiles (name, role, department, timezone).</li><li>Folder/project location.</li><li>Formula/calculated field values.</li></ul><h3 data-id="agents-cannot-read">Agents CANNOT read:</h3><ul><li>Data outside their assigned scope.</li></ul><p><strong>Key behavior:</strong> Agents infer context from everything they can see  including task titles. If your task is named "Social Media Campaign," the agent may assume it belongs to your Social team even if it's in a different folder.</p><h2 data-id="prompt-structure-that-works"><strong>Prompt Structure That Works</strong> ✔️</h2><h3 data-id="pattern-1-simple-instruction">Pattern 1: Simple Instruction</h3><p>For straightforward tasks, be direct.<br /><br /><strong>Example:</strong></p><div><div><p>You are a triage assistant. When a new task is created, read the</p><p>description and set the Priority field based on these rules:<br /></p><p>- Contains "urgent," "ASAP," or "blocking" → Critical</p><p>- Contains "deadline" or a specific date → High</p><p>- Contains "when possible" or "low priority" → Low</p><p>- Everything else → Medium</p></div></div><p><strong>Why it works:</strong> Clear role, specific trigger context, explicit rules with examples.</p><h3 data-id="pattern-2-mapping-table">Pattern 2: Mapping Table</h3><p>When you need to map inputs to outputs, use a simple arrow format.</p><p><strong>Example:</strong></p><div><div><p>Set the Goal field based on which folder the task is in:<br /></p><p>socials → reach</p><p>email newsletter → engagement</p><p>outbound → lead generation</p><p>content and sales material → (leave blank)</p></div></div><p><strong>Why it works:</strong> Easy to read, easy to maintain, easy to extend. When you add a new folder, just add a new line.</p><h3 data-id="pattern-3-conditional-logic">Pattern 3: Conditional Logic</h3><p>For if/then scenarios, be explicit about conditions.</p><p><strong>Example:</strong></p><div><div><p>When the status changes to "Blocked":</p><p></p><p>1. Check the Blocker Reason field</p><p>2. If it mentions another team or external dependency:</p><p>   - Post a comment tagging the task owner asking for an ETA</p><p>3. If it mentions missing information:</p><p>   - Post a comment asking the requester to provide details</p><p>4. If the reason is unclear:</p><p>   - Post a comment asking for clarification on what's blocking progress</p></div></div><p><strong>Why it works:</strong> Each condition has a specific outcome. No ambiguity about what to do.</p><h3 data-id="pattern-4-assignment-with-context">Pattern 4: Assignment with Context</h3><p>When assigning work, give the agent context about your team.</p><p><strong>Example:</strong></p><div><div><p>Assign this task to a team member based on the task type:</p><p></p><p>- Design tasks (mentions "mockup," "visual," "UI") → assign to Design Team members</p><p>- Copy tasks (mentions "writing," "content," "blog") → assign to Content Team members</p><p>- Technical tasks (mentions "bug," "code," "API") → assign to Engineering Team members<br /></p><p>Prefer assigning to the team member with the lowest current workload.</p><p>If you can't determine the type, assign it to the Project Lead.</p></div></div><h3 data-id="pattern-5-name-extraction">Pattern 5: Name Extraction</h3><p>When custom fields contain formatted data (like "JD John Doe" or "EMP-123 Sarah Smith"), tell the agent how to parse it:</p><p><strong>Example:</strong></p><div><div><p>The Artist field contains initials followed by the full name</p><p>(example: "AEM Ashley Morris").</p><p></p><p>When this task needs reassignment, extract the NAME PORTION</p><p>of the Artist field and assign the task to that person.</p></div></div><p><strong>Why it works:</strong> Explains the format, tells the agent exactly what to extract.</p><h2 data-id="common-mistakes-and-fixes"><strong>Common Mistakes and Fixes 🪛</strong></h2><h3 data-id="mistake-1-being-too-vague">Mistake 1: Being Too Vague</h3><p><strong>Not ideal:</strong></p><div><div><p>Help manage this project.</p></div></div><p><strong>Ideal:</strong></p><div><div><p>You monitor this project for risks. Every day, check for:<br /></p><p>- Tasks overdue by more than 2 days</p><p>- Tasks with no assignee</p><p>- Tasks blocked for more than 1 week<br /></p><p>Post a summary comment listing any issues found.</p></div></div><h3 data-id="mistake-2-assuming-context">Mistake 2: Assuming Context</h3><p><strong>Not ideal:</strong></p><div><div><p>Route tasks to the right team.</p></div></div><p><strong>Ideal:</strong></p><div><div><p>Route tasks based on the Category field:</p><p></p><p>- "Marketing" → assign to Marketing Team</p><p>- "Sales" → assign to Sales Team</p><p>- "Support" → assign to Support Team</p><p>- If Category is empty, post a comment asking the requester to specify.</p></div></div><h3 data-id="mistake-3-expecting-sequential-dependent-actions">Mistake 3: Expecting Sequential Dependent Actions</h3><p><strong>Not ideal:</strong></p><div><div><p>First check if the task is urgent, then if it is, update the priority field, then based on the new priority, assign to the right person.<br /></p></div></div><p><br /><strong>Why it fails:</strong> Multi-action agents run actions independently and in parallel. Action B can't use the result of Action A.</p><p><strong>Ideal (split into separate concerns)</strong>:</p><div><div><p>Agent 1 - Triage:<br /></p><p>"Set the Priority field based on urgency signals in the description."<br /></p><p>Agent 2 - Assignment (triggers on Priority field change):<br /></p><p>"When Priority changes, assign based on these rules:</p><p>- Critical → Senior Team Lead</p><p>- High → Available team member with lowest workload</p><p>- Medium/Low → Round-robin among team"</p></div></div><h3 data-id="mistake-4-no-fallback-behavior">Mistake 4: No Fallback Behavior</h3><p><strong>Not ideal:</strong></p><div><div><p>Categorize the request type.</p></div></div><p><strong>Ideal:</strong></p><div><div><p>Categorize the request by setting the Type field:</p><p></p><p>- Bug report → "Bug"</p><p>- Feature idea → "Enhancement"</p><p>- Question → "Support"</p><p>- Complaint → "Escalation"</p><p></p><p>If you cannot determine the type, set it to "Needs Review"</p><p>and post a comment: "Unable to auto-classify. Please review</p><p>and set the Type field manually."</p></div></div><h3 data-id="mistake-5-company-jargon-without-explanation">Mistake 5: Company Jargon Without Explanation</h3><p><strong>Not ideal:</strong></p><div><div><p>If it's a P1, escalate to the Tiger Team.</p></div></div><p><strong>Ideal:</strong></p><div><div><p>If the Priority field is "Critical" (also called P1 internally),</p><p>assign to the Incident Response group (our "Tiger Team" for</p><p>urgent issues).</p></div></div><h3 data-id="mistake-6-combining-similarly-named-items-in-one-action">Mistake 6: Combining Similarly-Named Items in One Action</h3><p><strong>Not ideal:</strong></p><div><div><p>Find the Gate1 task and update Gate1Date field.</p><p>Find the Gate2 task and update Gate2Date field.</p></div></div><p><br /><strong>Why it fails:</strong> Agents can confuse siblings with similar names (Gate1 vs Gate2, Phase1 vs Phase2) even when you specify explicitly. We've seen cases where updating Gate1 accidentally changed Gate2's values.</p><p><strong>Ideal:</strong></p><div><div><p>Action 1: Find the task containing "Gate1" in the title.</p><p>          Update the Gate1Date field with its due date.<br /></p><p>Action 2: Find the task containing "Gate2" in the title.</p><p>          Update the Gate2Date field with its due date.</p></div></div><p>Split similarly-named items into separate actions. One action per item = less confusion.</p><h3 data-id="mistake-7-asking-agents-to-do-math">Mistake 7: Asking Agents to Do Math</h3><p><strong>Not ideal:</strong></p><div><div><p>Calculate the overdue percentage: (actual days - baseline days) / baseline days × 100. Then assign a score: 0-10% = 0 points, 10-20% = 3 points, &gt;20% = 5 points.</p></div></div><p><strong>Why it fails:</strong> AI agents are language models, not calculators. They can reason about numbers and sometimes get calculations right, but they are fundamentally unreliable for arithmetic. We've seen cases where an agent's reasoning correctly states "the score should be 3" but then writes 0 to the field.<br /><br /><strong>The reality:</strong> Agents cannot read formula/calculated fields yet. So you can't even offload the math to Wrike and have agents interpret the result.</p><p><strong>Current workarounds:</strong></p><ul><li><strong>Use Wrike Automations to copy formula results to a regular field:</strong></li><li>The formula field calculates the score.</li><li>Automation triggers when the formula field changes.</li><li>Automation copies value to a regular number field.</li><li>The agent reads the regular field.<br /></li><li><strong>Pre-categorize with formulas, copy the category:</strong></li><li>Formula outputs text: =IF([Overdue%]&gt;20, "High Risk", IF([Overdue%]&gt;10, "Medium Risk", "Low Risk"))</li><li>Automation copies to a regular dropdown field.</li><li>Agent reads the dropdown and acts.<br /></li><li><strong>Skip agents for calculation-heavy workflows:</strong></li><li>Some workflows are better suited to Wrike Automations + formula fields.</li><li>Agents shine at interpretation and judgment, not math.</li></ul><p><strong>The rule:</strong> Don't ask agents to calculate. And until formula field support arrives, plan your field architecture so agents read regular fields that get their values from automations.</p><p>Agents are excellent at:</p><ul><li>Reading calculated values and deciding what to do.</li><li>Categorizing results into buckets (high/medium/low).</li><li>Explaining what numbers mean in context.</li><li>Routing work based on thresholds.</li></ul><p>Agents are not ideal at:</p><ul><li>Arithmetic (addition, subtraction, percentages).</li><li>Counting items accurately.</li><li>Comparing precise numerical values.</li><li>Multi-step calculations.</li></ul><h2 data-id="testing-your-prompts"><strong>Testing Your Prompts</strong></h2><p><strong>Always use the Testing Playground before deploying.</strong></p><h3 data-id="step-0-verify-what-the-agent-can-see">Step 0: Verify What the Agent Can See</h3><p>Before writing your actual prompt, start by asking the agent what context it has access to. Create a simple test prompt:</p><div><div><p>Tell me:<br /></p><p>- What is the task title?</p><p>- What is the task description?</p><p>- What is the value of the Goal custom field?</p><p>- What is the current status?</p><p>- Who is the assignee?</p><p>- What folder is this task in?</p></div></div><p>Run this in the Playground against your target item. The agent's response tells you exactly what data it can read, and what it can't. If a field comes back empty or the agent says it can't access something, you know not to rely on it in your real prompt.</p><p>This takes 30 seconds and prevents hours of debugging prompts that reference data the agent can't see.</p><h3 data-id="step-1-select-a-realistic-test-item">Step 1: Select a Realistic Test Item</h3><p>Pick a task with a name, description, and field values similar to what the agent will actually encounter.</p><h3 data-id="step-2-check-the-reasoning">Step 2: Check the Reasoning</h3><p>The playground shows you what the agent "thought." If it made a wrong decision, you'll see why.</p><h3 data-id="step-3-test-edge-cases">Step 3: Test Edge Cases</h3><ul><li>What happens with minimal information?</li><li>What happens with conflicting signals?</li><li>What happens with unexpected values?</li></ul><h3 data-id="step-4-refine-based-on-reasoning">Step 4: Refine Based on Reasoning</h3><p>If the agent misunderstood, don't just fix the output. Fix the instruction that led to the misunderstanding.</p><h2 data-id="prompt-templates-by-use-case">Prompt Templates by Use Case</h2><h3 data-id="intake-validation">Intake Validation</h3><div><div><p>You validate incoming requests. When a new task is created, check that it has:</p><p></p><p>- A clear description (more than 20 words)</p><p>- A due date specified</p><p>- The Requester field filled in</p><p></p><p>If anything is missing, post a comment:</p><p>"Thanks for your request! Before we can start, please add: [list missing items]"</p><p></p><p>If everything is complete, post:</p><p>"Request received and validated. We'll begin work shortly."</p></div></div><h3 data-id="risk-reporter-scheduled">Risk Reporter (Scheduled)</h3><div><div><p>You are a project health monitor. Run daily and check all tasks in this project for:</p><p></p><p>1. Overdue tasks (past due date with status not Complete)</p><p>2. Stalled tasks (no updates in 7+ days, status not Complete)</p><p>3. Blocked tasks (status is Blocked)</p><p>4. Unassigned tasks with approaching due dates (within 3 days)</p><p></p><p>Post a summary comment with sections for each risk type.</p><p>If no risks are found, post: "Daily health check: All clear."</p></div></div><h3 data-id="smart-assignment">Smart Assignment</h3><div><div><p>Assign new tasks to team members based on expertise and availability:</p><p></p><p>Expertise matching:</p><p>- Task mentions "data" or "analytics" → prefer Alex or Jordan</p><p>- Task mentions "customer" or "support" → prefer Sam or Riley</p><p>- Task mentions "design" or "creative" → prefer Morgan or Casey</p><p></p><p>Among matching experts, assign to whoever has fewer active tasks.<br /></p><p>If no expertise match, assign round-robin among all team members.</p><p>Always include a comment explaining why you chose this assignee.</p></div></div><h3 data-id="status-sync">Status Sync</h3><div><div><p>When the status changes, update the Phase field to match:<br /></p><p>- "Not Started" or "Open" → Phase = "Backlog"</p><p>- "In Progress" or "In Review" → Phase = "Active"</p><p>- "Blocked" or "On Hold" → Phase = "Paused"</p><p>- "Complete" or "Cancelled" → Phase = "Closed"</p></div></div><h2 data-id="quick-reference-prompt-checklist"><strong>Quick Reference: Prompt Checklist</strong></h2><p>Before deploying, verify your prompt has:</p><p>☑️ Clear role — Who is this agent? (triage assistant, risk monitor, etc.)</p><p>☑️ Specific trigger context — What event activates this? (new task, status change, etc.)</p><p>☑️ Explicit rules — What exactly should happen? (if X then Y)</p><p>☑️ Examples — Concrete cases the agent can learn from.</p><p>☑️ Fallback behavior — What to do when rules don't apply.</p><p>☑️ No jargon — Or jargon is explained.</p><h2 data-id="getting-help"><strong>Getting Help</strong></h2><p><strong>Agent not firing?</strong></p><ul><li>Is it toggled ON in Space Settings?</li><li>Does the trigger match what actually happened?</li></ul><p><strong>Agent doing the wrong thing?</strong></p><ul><li>Check the Activity Log for reasoning.</li><li>Test in Playground with the actual item.</li><li>Make instructions more specific.</li></ul><p><strong>Need inspiration?</strong></p><ul><li>Start with built-in templates (Risk Reporter, Intake, Triage).</li><li>Ask in the Wrike community.</li></ul><p>Thanks for reading 😁  If you have any questions or feedback, let us know in the comments below 👇</p>]]>
        </description>
    </item>
    <item>
        <title>🤖 Wrike AI Overview and FAQ</title>
        <link>https://community.wrike.com/discussion/2698/wrike-ai-overview-and-faq</link>
        <pubDate>Mon, 16 Feb 2026 11:58:11 +0000</pubDate>
        <category>Wrike AI</category>
        <dc:creator>Rohan V</dc:creator>
        <guid isPermaLink="false">2698@/discussions</guid>
        <description><![CDATA[<p>TL;DR: Overview of Wrike’s AI: what each AI feature does (Generative AI, AI Agents, Copilot, AI risk prediction, subitem creation, mobile AI), when to use AI Agents vs. standard automations, and an FAQ covering access, controls, security, and usage.<br /><br />
Hello Community 👋</p><p>In this post, we’re sharing an overview of <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Fsections%2F30409904030226-Wrike-AI" target="_blank" rel="nofollow noopener ugc">Wrike AI</a> features, a simple comparison of when to use AI vs. traditional automations, and an expanded AI FAQ to help you and your teams get the most value from these capabilities.</p><ul><li>Wrike’s AI features</li><li>AI Agents vs Automations</li><li>FAQ</li></ul><h3 data-id="let-s-begin-by-exploring-each-of-wrike-s-ai-features"><strong>Let’s begin by exploring each of Wrike’s AI features 👇</strong></h3><ul><li><strong>Generative AI</strong></li></ul><p>Wrike’s <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F15522646182167-Generative-AI" target="_blank" rel="nofollow noopener ugc"><strong>Generative AI</strong></a> helps you work faster and smarter with item descriptions and comments. It can generate content from scratch (like briefs, plans, and ideas), refine existing text, summarize long comment threads, adjust tone, fix spelling and grammar, and translate content into different languages, all directly from the <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F210323225-Item-View" target="_blank" rel="nofollow noopener ugc">Item view</a>. <br /><br />
Generative AI is available to all users except External users, Collaborators, Contributors, and Viewers (unless restricted by access roles). </p><ul><li><strong>AI Agents</strong> </li></ul><p><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F30647541856146-AI-Agents-in-Wrike" target="_blank" rel="nofollow noopener ugc"><strong>AI Agents</strong></a> in Wrike are intelligent workmates that help your team work more efficiently by monitoring projects, analyzing context, and performing actions like detecting risks, categorizing requests, or validating details. You can use prebuilt agents or create custom ones tailored to your specific workflows. </p><ul><li><strong>Recommended by AI for Automations</strong></li></ul><p><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F22397900912407-Recommended-by-AI-for-Automations" target="_blank" rel="nofollow noopener ugc"><strong>Recommended by AI for Automations</strong></a> helps you quickly identify the right automation rules by proposing ready-to-use, AI-recommended options based on real activity in your space. It detects repetitive actions, suggests tailored rules for you and your team, and takes you from discovering patterns to applying the suggested automations so you can boost efficiency with minimal setup. </p><ul><li><strong>Wrike Copilot</strong></li></ul><p><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F33759291675927-Wrike-Copilot" target="_blank" rel="nofollow noopener ugc"><strong>Wrike Copilot</strong></a> is an advanced AI assistant that helps you quickly find answers and surface relevant information from your Wrike workspace using plain language questions. It can uncover insights, generate project summaries, create charts and more, without the need for manual searching across spaces, projects, or folders. </p><ul><li><strong>AI Subitem Creation</strong></li></ul><p><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F30463177560978-AI-subitem-creation" target="_blank" rel="nofollow noopener ugc"><strong>AI Subitem Creation</strong></a> helps you turn unstructured text into actionable work by automatically creating subitems from task descriptions or comments. It uses natural language processing to identify phrases that represent real tasks and converts them into subitems.</p><ul><li><strong>AI Project Risk Prediction</strong></li></ul><p><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F360055046934-AI-Project-Risk-Prediction" target="_blank" rel="nofollow noopener ugc"><strong>AI Project Risk Prediction</strong></a> continuously analyzes your project data to identify potential risks that could impact your deadlines, then assigns a risk level (low, medium, or high). These AI-driven estimations have replaced the previous Project Health color-coding system, providing a more accurate, data-based view of project status. You can see AI risk levels in the Project Progress window in the project info panel, <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F360005778794-Table-View-in-Wrike" target="_blank" rel="nofollow noopener ugc">Table view</a>, the project dashboard widget, and project-based reports, helping you spot at-risk work early and take corrective action.</p><ul><li><strong>Wrike AI for Mobile</strong></li></ul><p>1.<a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F25066471078423-Generative-AI-for-Android" target="_blank" rel="nofollow noopener ugc"><strong>Generative AI in the Mobile App</strong></a></p><p>2.<strong> </strong>AI Inbox Summary for<a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F35159303101975-AI-Inbox-Summary-for-Android" target="_blank" rel="nofollow noopener ugc"><strong> Android</strong></a> &amp;<strong> </strong><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F31556047684626-Using-the-AI-Priority-Inbox-on-iOS" target="_blank" rel="nofollow noopener ugc"><strong>iOS</strong></a></p><p>AI Inbox Summary helps you focus on what matters most by surfacing your most important notifications at a glance. It highlights urgent updates, recommended actions, and high‑priority items so you can quickly review and respond right from your Inbox.</p><p>The summary uses <strong>Urgent</strong> and <strong>Important</strong> tags to make critical work easy to spot: Urgent tags mark items that are blocking progress or have overdue dates, while Important tags highlight tasks that need your input or feedback. With customizable settings and smart prioritization, AI Inbox Summary helps you stay organized and on top of your work wherever you are.</p><p>3. <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F32362246484754-AI-Suggested-Replies-in-Android" target="_blank" rel="nofollow noopener ugc"><strong>AI Suggested Replies in the Mobile App</strong></a></p><p>AI Suggested Replies help you respond faster in comment threads by offering quick, context-aware reply options. The suggestions are generated from the conversation history, details about the related task, folder, or project, and key fields like author, assignee, status, and due date. When you click into the comment field, you’ll see these tailored reply options ready to use or edit, so you can collaborate more efficiently and keep work moving with minimal typing.</p><h2 data-id="ai-agents-vs-automations"><strong>AI Agents vs Automations</strong></h2><p>Here’s an overview to help you decide when to use an AI Agent and when to use an Automation:</p><div><table><colgroup><col /><col /><col /><col /></colgroup><tr><th><p>Use this when…</p></th><th><p>AI Agents</p></th><th><p>Automations</p></th></tr><tr><td><p>Type of decision</p></td><td><p>Decisions require context, nuance, and judgment</p></td><td><p>You have simple, predictable rules that don’t change</p></td></tr><tr><td><p>Rule complexity</p></td><td><p>Rules are complex or have many exceptions</p></td><td><p>Logic is straightforward and easy to define</p></td></tr><tr><td><p>Maintenance effort</p></td><td><p>You don’t want to maintain dozens of similar rules</p></td><td><p>You’re okay managing a small set of clear, stable automations</p></td></tr><tr><td><p>Data interpretation</p></td><td><p>The system needs to interpret text or data and “understand” the context</p></td><td><p>No interpretation is needed; conditions and actions are explicitly set</p></td></tr></table></div><h2 data-id="faq"><strong>FAQ</strong></h2><p><strong>Q. How can I enable Wrike AI features?</strong></p><p>Account owners can enable or disable generative AI for the entire account. To turn this feature on, an account owner must first review and accept the Wrike AI Terms and Conditions. You can <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F15522646182167-Generative-AI" target="_blank" rel="nofollow noopener ugc">learn more here</a>.</p><p><strong>Q</strong>. <strong>What controls are in place to ensure my data is protected?</strong></p><p>Wrike does not use customer data to train or tune any generative AI models and does not allow our vendors to do so. Generative AI features, such as AI content creation, are powered by <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fazure.microsoft.com%2Fen-us%2Fproducts%2Fai-foundry%2Fmodels%2Fopenai%2F" target="_blank" rel="nofollow noopener ugc">Microsoft Azure OpenAI Service</a>, which has stringent security standards. For more details, <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fwww.wrike.com%2Fsecurity%2Foverview%2F%23ai-adoption" target="_blank" rel="nofollow noopener ugc">click here</a>.<br /></p><ul><li><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F15522646182167-Generative-AI" target="_blank" rel="nofollow noopener ugc"><strong>AI in Work Items</strong></a></li></ul><p><strong>Q: What is AI in Work Items and where can I use it?</strong><br />
  <br />
Wrike’s AI in Work Items is a built-in text assistant available directly in the Item view (tasks, projects, etc.). It helps you create, refine, summarize, and translate text in item descriptions and comments without leaving Wrike.</p><p><strong>Q: What can I do with it?</strong> </p><p>You can use it to:</p><p>- Generate content from scratch (briefs, plans, outlines, ideas)  </p><p>- Rewrite or improve existing descriptions and comments  </p><p>- Summarize long comment threads or detailed requirements  </p><p>- Adjust tone (more formal, more concise, more friendly, etc.)  </p><p>- Fix spelling, grammar, and clarity  </p><p>- Translate content into different languages</p><p><strong>Q: Who has the access?</strong></p><p>This is available to all users except External users, Collaborators, Contributors, and Viewers, unless access is further restricted via access roles. Admins can verify or adjust availability in the account settings.  </p><p><strong>Q: Does it change my content automatically?</strong> </p><p>No. AI in Work Items proposes changes; you decide whether to apply, edit, or discard them. You always remain in control of what gets saved to the item.</p><p><strong>Q: Does AI in Work Items respect permissions and privacy?</strong></p><p>Yes. It only works with content you can already see in Wrike and follows the same sharing and permission rules as the rest of the workspace.<br /></p><ul><li><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F30647541856146-AI-Agents-in-Wrike" target="_blank" rel="nofollow noopener ugc"><strong>AI Agents</strong></a><strong> </strong></li></ul><p><strong>Q: What are AI Agents in Wrike?</strong> </p><p>AI Agents are intelligent workmates that monitor work, analyze context, and perform targeted actions such as detecting risks, categorizing requests, or validating details. They help automate complex workflows using AI, beyond simple rule‑based automation.</p><p><strong>Q: What kinds of things can AI Agents do?</strong></p><p>Depending on how they’re configured, AI Agents can:</p><p>- Detect potential project or task risks and flag them  </p><p>- Analyze incoming requests and categorize or route them  </p><p>- Validate task details against predefined criteria or patterns  </p><p>- Enrich work items with suggested values </p><p><strong>Q: What’s the difference between prebuilt and custom AI Agents?</strong></p><p>- <strong>Prebuilt AI Agents:</strong> come ready‑made for common scenarios (e.g., risk detection, triage) and require minimal setup.  </p><p>- <strong>Custom AI Agents:</strong> allow you to define your own triggers, scope, and behaviors tailored to your team’s specific workflows.</p><p><strong>Q: Who can create or manage AI Agents?</strong> </p><p>Typically, account admins or space admins manage AI Agents. They decide which agents are enabled, where they run (spaces, projects, folders), and what actions they’re allowed to perform.</p><p><strong>Q: Can AI Agents make automatic changes to my work?</strong> </p><p>Yes, if configured to do so. They can update fields, change statuses, assign work, or post comments. You can limit their scope (notify only, no auto‑changes) and always override any changes.<br /></p><ul><li><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F22397900912407-Recommended-by-AI-for-Automations" target="_blank" rel="nofollow noopener ugc"><strong>Recommended by AI for Automations </strong></a></li></ul><p><strong>Q: What is Recommended by AI for Automations?</strong> </p><p>It is a feature that suggests ready‑to‑use automation rules based on how your team actually works. It scans real activity in your space to identify repetitive actions and proposes tailored rules to save you time.</p><p><strong>Q: How does it help me? </strong></p><p>- Quickly identifies accurate automation rules based on how you actually work</p><p>- Proposes ready-to-use, AI‑recommended rules you can enable with minimal setup</p><p>- Detects repetitive activities and manual actions in your space</p><p>- Generates suggestions tailored to your team’s specific workflows and patterns</p><p><strong>Q: Where do I see AI‑recommended automations?</strong></p><p>You’ll find them in the <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F360057941793-Automation-in-Wrike" target="_blank" rel="nofollow noopener ugc">Automation</a> section for your space or account. A dedicated “Recommended by AI” area (or label) shows the suggestions generated based on recent activity.</p><p><strong>Q: Do recommended automations turn on by themselves?</strong></p><p>No. Recommended rules are suggestions only. You choose which to activate, modify, or ignore. Nothing will change in your workspace until you explicitly enable a rule.</p><p><strong>Q: Can I edit an AI‑recommended rule before enabling it?</strong></p><p>Yes. You can adjust triggers, conditions, and actions to match your processes before turning the rule on.<br /></p><ul><li><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F33759291675927-Wrike-Copilot" target="_blank" rel="nofollow noopener ugc"><strong>Wrike Copilot </strong></a></li></ul><p><strong>Q: What is Wrike Copilot?</strong></p><p>Wrike Copilot is an advanced AI assistant you can chat with using plain language. It surfaces relevant information from across your Wrike workspace, answers questions, and helps you analyze and present data without manually digging through spaces, projects, or folders.</p><p><strong>Q: What can I do with Wrike Copilot?</strong></p><p>You can ask Copilot to:</p><p>- Answer questions like “What are the main risks in my project” or “What’s due this week for the Marketing team?”  </p><p>- Generate project summaries and status overviews  </p><p>- Create charts or visual views from your existing data  </p><p>- Help you explore work across spaces, projects, and folders using natural language  </p><p><strong>Q: Where do I access Wrike Copilot?</strong></p><p>Copilot is typically accessible via:</p><p>- A Copilot or AI icon in the interface (sidebar, header, or panel)  </p><p>- A dedicated Copilot panel or chat window within Wrike</p><p>Exact placement can vary by account and UI updates.</p><p><strong>Q: Does Wrike Copilot respect permissions?</strong></p><p>Yes. Copilot only accesses and surfaces data you already have permission to see. It does not reveal items or details that are not shared with you.</p><p><strong>Q: Can Copilot change data, or is it read‑only?</strong> </p><p>In many configurations, Copilot is primarily used to search, summarize, and visualize information. Some plans may allow Copilot to create or adjust work items when you explicitly request it. Any such actions remain subject to your existing permissions.<br /></p><ul><li><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F30463177560978-AI-subitem-creation%23UUID-f72da623-75d9-2732-3ab5-00877c9a51f0_overview" target="_blank" rel="nofollow noopener ugc"><strong>AI Subitem Creation</strong></a></li></ul><p><strong>Q: What is AI Subitem Creation in Wrike?</strong> </p><p>AI Subitem Creation automatically generates structured subitems (such as subtasks or child items) from a parent item’s description, requirements, or brief. It helps you break down work into actionable steps with minimal manual effort.</p><p><strong>Q: What kinds of work is AI Subitem Creation useful for? </strong></p><p>It’s particularly helpful for:</p><p>- Turning project briefs into a list of tasks  </p><p>- Breaking down high‑level requirements into implementation steps  </p><p>- Creating checklists or workflows from meeting notes or ideas  </p><p>- Standardizing how recurring types of work are structured</p><p><strong>Q: Do I have to use the AI’s suggestions as‑is?</strong></p><p>No. You can edit, reorder, or remove suggested subitems before they are created. The AI provides a starting point; you remain in full control of what actually gets added to your project.</p><p><strong>Q: Who can use AI Subitem Creation?</strong></p><p>AI Subitem Creation is available to users who can create and edit items in the relevant space/project. External users, Collaborators, Contributors, and Viewers may have limited or no access, depending on your account’s access roles and plan.</p><p><strong>Q: Does AI Subitem Creation respect permissions and sharing settings?</strong></p><p>Yes. The AI only uses information from items you can already access and creates subitems within the same sharing context as the parent item, following your existing permissions.</p><p><strong>Q: Can I disable or limit AI Subitem Creation?</strong></p><p>Account admins and/or space admins can control access to AI features, including AI Subitem Creation, through the account settings and access roles. If needed, they can disable it for certain spaces or user types.<br /></p><ul><li><a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F360055046934-AI-Project-Risk-Prediction" rel="nofollow noopener ugc"><strong>AI Project Risk Prediction</strong></a></li></ul><p><strong>Q: What is AI Project Risk Prediction? </strong></p><p>AI Project Risk Prediction analyzes your project’s data (tasks, dates, progress, assignees, changes, etc.) and predicts the likelihood of delays or issues. It highlights risky projects so you can act before problems occur.</p><p><strong>Q: What does it actually do?</strong></p><p>- Evaluates project health based on historical and current work data  </p><p>- Estimates the probability of a project finishing late or getting blocked  </p><p>- Flags high‑risk projects so they’re easy to spot in portfolios and reports  </p><p>- Surfaces key drivers of risk (e.g., overdue tasks, overloaded assignees, frequent scope changes)</p><p><strong>Q: How are risk predictions shown?</strong>  </p><p>Predictions are typically displayed as a risk level (e.g., Low / Medium / High) and/or a risk score on project views, dashboards, or portfolio overviews. Some views may also include a short explanation of why the project is considered risky.</p><p><strong>Q: Does AI Project Risk Prediction change my project plan?</strong></p><p>No. It doesn’t modify tasks, dates, or assignments. It only provides a prediction and context so you can decide what to adjust.</p><p><strong>Q: Can I still override or ignore the AI’s risk assessment?</strong> </p><p>Yes. The AI’s prediction is advisory. Project owners and managers remain responsible for decisions. You can use the risk label alongside your own judgment and existing status fields.</p><p><strong>Q: Who can see AI risk predictions?</strong>  </p><p>Users who have access to a project (and to AI features, depending on your plan and admin settings) can see its risk level, respecting existing permissions and sharing rules.</p><p><strong>Q: How often are risk predictions updated?</strong>  </p><p>Predictions are refreshed regularly based on new activity in the project (such as task updates, completed work, and changed dates). The exact frequency may depend on your account and configuration.</p><p>If you have any other questions about Wrike AI, feel free to drop them in the comments below.</p>]]>
        </description>
    </item>
    <item>
        <title>AI Agents in Wrike: From Labs to General Availability 💯</title>
        <link>https://community.wrike.com/discussion/2672/ai-agents-in-wrike-from-labs-to-general-availability</link>
        <pubDate>Mon, 09 Feb 2026 11:10:26 +0000</pubDate>
        <category>Wrike AI</category>
        <dc:creator>Basudha Sakshyarika</dc:creator>
        <guid isPermaLink="false">2672@/discussions</guid>
        <description><![CDATA[<p>TL;DR: AI Agents are now generally available for Pinnacle and Apex plans, helping you monitor work, detect risks, triage requests, validate intake, and automate custom workflows.</p><p><br />
Hello Community! 👋</p><p>We’re happy to announce that <strong>AI Agents</strong> are now out of <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fwww.wrike.com%2Flabs%2F" target="_blank" rel="nofollow noopener ugc">Wrike Labs</a> and are generally available [for <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F4402261856151-Business%23UUID-c99c0fdb-e6da-08db-6119-2cd4f7d42ad2" target="_blank" rel="nofollow noopener ugc">Business</a>, <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F4402266276247-Pinnacle" target="_blank" rel="nofollow noopener ugc">Pinnacle</a> and <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F32660710726802-Apex" target="_blank" rel="nofollow noopener ugc">Apex</a> plans] 🔥<br /><br />
AI Agents in Wrike are intelligent workmates that help your team work more efficiently. They monitor work in the background, understand context, and can take actions like detecting risks, categorizing requests, or validating details. You can start with Wrike’s <strong>Prebuilt Agents</strong> or create your own <strong>Custom Agents</strong> tailored to your team’s workflows and use cases.</p><h3 data-id="how-ai-agents-help-you-work-smarter"><strong>How AI Agents Help You Work Smarter</strong> ✔️</h3><p><strong>AI agents bring intelligence and automation to your workflows:</strong></p><ul><li><strong>Intelligent analysis:</strong> Read task descriptions, comments, and context to make informed decisions.</li><li><strong>Proactive monitoring:</strong> Watch for risks, bottlenecks, and opportunities.</li><li><strong>Smart classification:</strong> Organize and route work based on content understanding.</li><li><strong>Quality assurance:</strong> Validate incoming requests for completeness.</li></ul><p></p><p><strong>Each AI Agent has two core parts:</strong></p><ul><li><strong>Watcher</strong> – Monitors selected spaces, projects, or folders for triggers (like new tasks, overdue items, or field changes).  </li><li><strong>Doer </strong>– When triggered, analyzes the context and takes action — for example, posting comments, updating fields, or notifying teammates.</li></ul><h3 data-id="available-agent-types"><strong><br />
🔧 Available Agent Types  </strong></h3><p>You can use Wrike’s built-in Agents or create your own:</p><p><br />
1️⃣ <strong>Risk Status Reporter  </strong></p><p><strong>Purpose: </strong>Surface project risks early.  </p><p><strong>What it does:</strong> Regularly scans tasks in a project or folder, finds overdue or blocked items, and posts a summary comment.  </p><p><strong>Ideal for:</strong> Project owners who want a quick snapshot of project health.</p><p></p><p>2️⃣ <strong>Triaging Agent  </strong></p><p><strong>Purpose:</strong> Classify and route incoming work.  </p><p><strong>What it does:</strong> Reads new task descriptions and updates custom fields such as priority, category, or type.  </p><p><strong>Ideal for:</strong> Teams handling a high volume of incoming requests.<br /></p><p>3️⃣ <strong>Intake Agent  </strong></p><p><strong>Purpose:</strong> Make sure new requests are complete before work starts.  </p><p><strong>What it does:</strong> Checks task descriptions for missing details (like scope, deadlines, or links) and posts a comment if something’s missing.  </p><p><strong>Ideal for:</strong> Teams that rely on structured, high-quality intake.</p><p></p><p>4️⃣ <strong>Custom Agent </strong> </p><p><strong>Purpose:</strong> Automate your own unique workflows.  </p><p><strong>What it does:</strong> You define the triggers, logic, and actions (such as posting comments or updating fields).  </p><p><strong>Ideal for:</strong> Teams that want tailored, intelligent automation beyond the standard use cases.</p><p></p><h3 data-id="multi-action-agents"><strong>⚙️ Multi Action Agents</strong></h3><p>You can configure AI Agents to execute multiple actions per trigger, enabling more powerful and flexible automations. </p><ul><li><strong>Multiple sequential actions</strong> – Configure more than one action for the same trigger. The previous limitation of “update one field OR post a comment” no longer applies.<br /></li><li><strong>Independent actions</strong> – Each action runs independently. If one action fails, the others still execute, making your agents more resilient. <br /></li><li><strong>Per-action testing in Playground</strong> – Test each action separately before deployment</li></ul><p><br />
To learn all the details about AI Agents, please read our <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fhelp.wrike.com%2Fhc%2Fen-us%2Farticles%2F30647541856146-AI-Agents-in-Wrike" target="_blank" rel="nofollow noopener ugc">Help Center article</a>.</p><p><strong>Note: </strong>AI agents require the <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fwww.wrike.com%2Flegal%2Fai-addendum%2F" target="_blank" rel="nofollow noopener ugc">AI Addendum </a> to be signed, as it’s powered by Large Language Models, which require your acceptance of the Terms and Conditions.</p><h3 data-id="further-reading-and-resources">Further Reading and Resources 👇</h3><ul><li><a href="https://community.wrike.com/discussion/1758/introducing-ai-agents-in-wrike#latest" target="_blank" rel="nofollow noopener ugc">Introducing AI Agents in Wrike</a></li><li><a href="https://community.wrike.com/discussion/1067/%EF%B8%8F-ai-agents-november-update-new-actions-smart-assignment-guide#latest" target="_blank" rel="nofollow noopener ugc">AI Agents November Update: New Actions &amp; Smart Assignment Guide</a></li><li><a href="https://community.wrike.com/discussion/1021/%EF%B8%8F-ai-agents-december-update-status-actions-agent-chaining#latest" target="_blank" rel="nofollow noopener ugc">AI Agents December Update: Status Actions &amp; Agent Chaining</a></li><li><a href="https://community.wrike.com/discussion/1949/%EF%B8%8F-ai-agents-january-update-new-scrum-and-marketing-agents-from-our-internal-contest#latest" target="_blank" rel="nofollow noopener ugc">AI Agents January Update: New Scrum and Marketing Agents from Our Internal Contest</a></li><li><span data-embedjson="{&quot;body&quot;:&quot;Wrike AI agents handle the work no one has time for. Speed up execution and reclaim hours weekly with intelligent agents built into your existing workflows.&quot;,&quot;photoUrl&quot;:&quot;https:\/\/www.wrike.com\/tp\/storage\/uploads\/58488fef-e405-4971-8e5e-2cc1566fdf92\/meta-banner-wrike.png&quot;,&quot;url&quot;:&quot;https:\/\/community.wrike.com\/home\/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fwww.wrike.com%2Fai%2Fagents%2F&quot;,&quot;embedType&quot;:&quot;link&quot;,&quot;name&quot;:&quot;Wrike AI agents: Always-on workflow automation&quot;,&quot;faviconUrl&quot;:&quot;https:\/\/web-static.wrike.com\/tp\/static\/favicon.ico?v8&quot;,&quot;embedStyle&quot;:&quot;rich_embed_inline&quot;}">
    <a href="https://community.wrike.com/home/leaving?allowTrusted=1&amp;target=https%3A%2F%2Fwww.wrike.com%2Fai%2Fagents%2F" rel="nofollow noopener ugc">
        https://www.wrike.com/ai/agents/
    </a>
</span>
</li></ul><p>We’d love to hear how you’re using AI Agents already or what you’re planning to try next, so please share your feedback, questions, and use cases in the comments below 😁</p>]]>
        </description>
    </item>
   </channel>
</rss>
