Seeking advice: Wrike AI rollout and governance

Question

Hi everyone,

We're currently exploring the implementation of Wrike AI within our organisation and I'd be really interested to hear how others approached the rollout, particularly from a governance and data management perspective.

Some of the questions we're considering are:

* How did you assess the risks and benefits before enabling Wrike AI?
* Did you put any governance or approval processes in place before making it available to users?
* How did you manage concerns around organisational data being used by AI?
* What guidance or training did you provide to users on appropriate AI usage?
* Did you restrict any AI features initially, or enable everything from the start?
* Were there any particular security, compliance, legal, or data protection considerations that influenced your approach?
* How did you communicate what data the AI could access and how it would (or wouldn't) be used?
* Looking back, is there anything you would do differently?

We're a higher education institution, so we're particularly interested in balancing the benefits of AI with appropriate governance, information security, and responsible use. We're developing our own internal approach and would really appreciate hearing about real-world experiences from other organisations.

Thanks in advance for sharing any lessons learned, recommendations, or documentation you're willing to discuss!

jtyler81 · Answer

These are great questions, Jessica. One approach I'd recommend is starting with a small pilot group made up of representatives from IT, Security, Legal/Compliance, and a few business users. That gives you the opportunity to validate governance, develop internal best practices, and identify any gaps before rolling AI out more broadly. I'd also document clear guidelines around which use cases are encouraged, which require review, and which should be avoided. Having that foundation early can make organization-wide adoption much smoother.

Ali Moses · Answer

When deciding if we should enable AI in Wrike, I partnered with our IT department. Luckily, Wrike has a lot of readily available documentation detailing what aspects of our data the AI would have access to and where it would be shared. (Nowhere.) This, and the fact that Wrike doesn't use any of our data to train their models helped us feel secure enough to allow AI usage in our account.

Our company's knowledge base is housed in Wrike, so for training and governance, I was able to post several articles there for our users to reference as well as promoting proper usage through our Wrike Teams channels.

At this time, only our space admins are able to create agents and I have asked that they not do so without consulting me. This is to make sure we do not go over our monthly action allowances.

John_Egleston · Answer

Great questions, Jessica! As an internal Wrike product manager as my previous nonprofit organization was evaluating various AIs, I got involved by raising the specific request of a project manager to our IT operations manager, who also held Wrike's "owner" role for the account. Giving them Wrike's documentation was helpful context for a provisional approval.

* How did you assess the risks and benefits before enabling Wrike AI? - It helped that the core Wrike AI features obeyed users' security as my Technology and Legal colleagues weighed in. (We weren't yet looking at MCP.)
* Did you put any governance or approval processes in place before making it available to users? - No, just enabled it and promoted it through an internal channel, letting adoption take an organic path.
* How did you manage concerns around organisational data being used by AI? - Since the data was already in Wrike and would stay in Wrike's AI, there was no problem with that.
* What guidance or training did you provide to users on appropriate AI usage? - I posted some examples and links to Wrike's help articles, encouraging experimentation and as-needed adoption and use.
* Did you restrict any AI features initially, or enable everything from the start? - No restrictions other than Wrike's account-wide usage limits, which I monitored. 
* Were there any particular security, compliance, legal, or data protection considerations that influenced your approach? - Not for the core features, except as noted above. MCP would definitely be another level, but I was just starting to experiment with agents at that stage.
* How did you communicate what data the AI could access and how it would (or wouldn't) be used? - I left that to the documentation I linked to.
* Looking back, is there anything you would do differently? - I would have engaged more with the leaders who were thinking more broadly about AI in the organization.

Glad to interact more about any of this!