Cognition's cloud engineer takes tickets from Slack, Linear or Jira and opens pull requests; free tier, Pro $20 a month, Max $200 a month.
Teams that want to delegate whole tickets rather than drive an editor: Devin picks up Linear or Jira work, reproduces and fixes bugs, builds internal tools, runs and tests its changes, and leaves draft pull requests waiting for review. Delegation happens where the team already talks, by tagging it in a Slack or Microsoft Teams thread, or by commenting /devin on a pull request. There is an embedded IDE for following along and taking over mid-session, plus a terminal CLI that hands longer work off to cloud Devin.
Developers who want a tight, local, editor-first loop will find this the wrong shape — Devin is a cloud teammate you brief and review, not an autocomplete. The vendor's own guidance also caps ambition: tasks a person could do in 'three hours or less' are likely feasible, longer work should be split into smaller sessions, and repeatedly hitting usage limits is treated as a sign the task is too complex. Anyone with a strict no-training policy must act, not assume: the security docs state that by default Cognition may use your data for model training, opting out requires a paid plan, and on the Teams plan only an administrator can do it. Hosting regions are not published.
Pros
Cons
There is a $0 Free plan with a light quota, then Pro at $20 a month and Max at $200 a month with significantly higher quotas. Teams is $80 a month for the team plan plus $40 a month per full dev seat, supporting up to 200 users, and Enterprise is custom quoted. Paid plans can also purchase extra usage at API pricing once the refreshing allowance runs out.
By default it may. Cognition's security docs state that by default it may use your data for model training, and that paid-plan users can opt out on the Data Controls settings page, with only an administrator able to do so on the Teams plan. Once you opt out, Cognition says training stops and Zero Data Retention is enabled with its model providers. Enterprise data requires prior written consent.
Devin creates pull requests on GitHub and Bitbucket and works with GitLab merge requests, and the docs describe a workflow of draft PRs waiting for review. The vendor frames the ideal outcome as you seeing CI pass and approving the PR, so review is reduced rather than removed. Devin Review and Auto-Fix exist to cut that review effort, not to eliminate the approval step.
Cognition's own guidance is specific: tasks a person could finish in three hours or less are likely feasible, and longer work should be split into smaller sessions. It says results are best where there are test suites, CI checks or other verifiable outcomes, and that success rates rise with clearer, more specific instructions. Repeatedly hitting usage limits is treated as a signal the task is too complex.
Facts sourced from the vendor’s own site, last verified 2026-10-10.
The Weekly Source
One email a week. New listings, repriced tools, and the ones we would skip - including this one if its pricing changes.
