Is rovo really usable for any enterprise ?

[Martin Poirier]

Hello,

We are currently evaluating Atlassian Rovo as part of our Data Center to Cloud migration. We are concerned about the sustainability of the usage-based credit model at an enterprise scale. Atlassian has been been pushing moving to cloud to use Rovo everywhere but here are some examples that are very limited.

We have currenctly 31500 credits for 15k users

Automation with rovo can bust your credit quickly

- An automation rovo to automatea component and labels selection with custom agent in test for 1 day with 1 person took 3000 credits already

Action Type	Example	Credits per Execution
Basic	Add label on issue creation	0
Simple	Extract keywords from summary and add as labels	10
Deep	Generate root cause analysis from all fields and attachments	100

250000 credit just for one automation

• Credit by execution : 10
• Total credits/month: 25000 (execution conservative for us, more on data center currently) x 10 = 250,000 credits

Deep searches are totally unusable on big sites

• Example : calculate a number field by platform field, 100 issues take about 10 minutes, 1000 issues never finish.  Our needs are easily more than 10k+.

Other AI does not charge for internal product simple chat

We have Microsoft copilot enterprise has no credit limit on internal microsoft app for chat with where rovo use credit for everything even their own app.    https://learn.microsoft.com/en-us/microsoft-copilot-studio/requirements-messages-management

In conclusion, we are not sure if pushing Rovo internally will be solution for us, since there's no clear direction on how we can stop the usage of Rovo on a site, an automation, or simple chat by users that can directly bust your montly credit for the whole enterprise in basicly 5 minutes.

Martin Poirier

Ubisoft

[Nicolas Bourdages _Ubisoft_]

It looks like it would be very easy to exceed available budgets, and there are no clear guardrails to limit credit spending per user or per group. It's nearly impossible to correlate AI credit spending with specific automations, Agents or users. I don't know how we, as admins, would be able to properly monitor usage, keep spending under control and identify problematic implementations.

If we can't leaving our users free to explore and come up with AI-powered solutions to the problems they have without risking them racking up large AI bills, what would be the point to even let them use Rovo?