01 / what I build
AI agents that take repetitive QA and bug-fixing work off your
engineers, running safely inside the CI and deploy setup you already
have. These three come up most often. Each one is something I have
built and written up in detail.
01
Higher-Confidence QA Gates
Every release waits on someone experienced to judge whether a
failed test is a real problem or a false alarm. That manual check
slows shipping and ties up the people you can least spare. Skip
it, and you ship on a guess. This agent makes the call for you,
the way your best engineer would, so the whole team gets a release
decision they can trust.
- Releases stop waiting on manual review of failed tests
- Tells a real problem from a false alarm, reliably enough to act on
- Frees your most experienced people from sifting through failures
- A release decision the whole team trusts, in about a minute
see related blog →
02
Manual Testing on Every Release, Automated
Every release needs someone to look it over for the problems
automated tests can't catch: an odd input, a broken permission, a
path no script covers. Done by hand that is hours of work on every
deploy, and shipping waits on it. This agent runs that same check
automatically on every release and reports back in minutes, so the
team gets a read on each deploy without waiting on anyone.
- Runs the hands-on check a manual tester would, on every release
- Catches the problems scripted and regression tests can't
- Reads what each release changed and focuses its testing there
- A clear report in about ten minutes, not hours by hand
see related blog →
03
Turn Bug Tickets Into Ready PRs
Hand it a bug report and it does the first pass of the fix:
reproduces the problem, writes the change, confirms nothing else
broke, and opens a pull request for your team to review. Every
step has a safety check, so it stops cleanly instead of guessing,
and it never merges on its own. Your engineers review a finished
draft instead of starting from a blank page.
- Turns a bug ticket into a ready-to-review pull request
- Reproduces the bug, then confirms the fix actually works
- Re-runs your test suite to check nothing else broke
- Always opens a PR for a human to review, never merges itself
see related blog →
02 / how I work
No template. We start from your actual pain and prove it on real work before it touches anything that matters.
-
Start with your pain.
Which one is loudest: red builds eating your team's time, bugs slipping past your tests, or fixes piling up? We pick the agent that helps most.
-
Build it for your setup.
I wire the agent into the CI and deploy flow you already use, matched to your stack and conventions. No new platform to adopt.
-
Keep humans in control.
Every agent reports in plain language and stops cleanly when it is unsure. Nothing ships without your team's review.
-
Expand on trust.
Once one agent earns its place, the others slot onto the same foundation.