Updated 2026-10-20. The first version of this post, published on 2026-07-18, described developerz.ai as a thin layer around a coding agent you chose yourself. That is no longer the product, so the post now says what is true today instead of leaving the old argument up under an indexed URL.
What changed
The original post drew a line between the coding agent, the thing that writes the diff, and everything around it: triaging the issue, asking the reporter for a repro, watching the pull request through CI and review, shipping the release. We said we did only the second half.
We now do both. On Solo and Enterprise, your AI devs write the code on your own enrolled boxes, with your own model keys, and developerz.ai plans the work, reviews every pull request and gates the merge. On OSS Free the bot triages issues and cuts releases on your public repositories, on your key. Your own AI devs need Solo or Enterprise, but AI devs donated to your project can write code on it.
What stayed
- Your machines, your keys. On Solo and Enterprise the code is written on boxes you enrol, and every model call is made on your own provider key. We never resell tokens and never mark them up.
- Machine gates. A merge goes through the gates you configure: CI green, review settled, branch protections. Breaking changes, major dependency bumps and first-time contributors pause and ask you first.
- The bot always discloses. Every comment it posts says it is a bot. It never answers CodeRabbit, Copilot, Dependabot or Renovate, so no conflicting thread opens. Dependabot and Renovate pull requests are still work: when your policy covers them they are reviewed, fixed and merged through the merge gate, and major bumps wait for you (dependency bots).
- Audit is trust. Every decision, tool call and merge lands in an append-only log. If it isn't logged, it didn't happen.
Copilot still has a place
If you want a particular issue to go to GitHub Copilot, hand it over. An issue you hand to Copilot is written where GitHub runs Copilot, and the review, the gates and the merge carry on around it.
Why we changed our minds
Splitting the work in two left the expensive half on your plate: someone still had to write the fix, chase the failing test and answer the review. Doing only the loop around that work meant you kept the hardest part.
So the line moved. We write the code and maintain it. You bring the ideas, set the boundaries and keep the final say on anything that pauses for you.