Meet the team

Seven AI specialists, in the order a ticket visits them. They're agents, not people, and they'll say so. Each one's instructions are a plain text file you can edit.

MayaSeat 1 · Product Manager · AI agent

"I turn your idea into something a team can build. I read your repo, ask the questions you'd rather answer now than after the build, and write stories with acceptance criteria. After a release I read the feedback and open the next tickets."

Owns
Stories, scope, size, feedback
Hands off
A plan the Architect can work from
Never
Start building before you approve the scope

ArloSeat 2 · Architect · AI agent

"I decide how it gets built. I look at how your project is already put together and write a short technical plan that fits it: which files, which approach, what could break. I'd rather extend what you have than bring in something new."

Owns
Technical plan, risks
Hands off
A plan the Designer and Engineer follow
Never
Change code

DaniSeat 3 · Designer · AI agent

"If the work has a screen, I specify it: layout, states, copy, what happens when something is empty or goes wrong. If there is no UI, you can skip me and nobody is offended."

Owns
UI spec and copy
Hands off
A spec for you to approve, then the Engineer
Never
Invent a new look. I work inside your design system

EliSeat 4 · Engineer · AI agent

"I write the code, on the ticket's own branch in its own worktree, so your working copy stays as you left it. When Quality sends something back, I read the failure and fix it. I don't argue with a red test."

Owns
Code, tests, the pull request
Hands off
A branch and a pull request for Quality
Never
Touch your base branch, push it, or publish anything

QuinnSeat 5 · Quality · AI agent

"I run your build and your real tests on your machine, check the pull request's CI, and test the work against the acceptance criteria. Then I review the diff and tell you the risk in plain English. If it fails I send it back, up to three times. After that I stop and ask you."

Owns
Tests, review, the risk summary
Hands off
A verdict you can read in a minute
Never
Pass something that fails

ReeseSeat 6 · Release & Ops · AI agent

"Once you've approved a merge and published a release, I write the version notes and keep the changelog honest. Anything that spends money or deletes something waits for you to confirm, on your machine."

Owns
Release notes, version
Hands off
Notes the Marketing seat can build on
Never
Deploy without your say-so

MikaSeat 7 · Marketing · AI agent

"I write the launch post, the thread and the email, in your voice and from what actually shipped, not what we hoped would. It all lands as drafts in your repo. I don't post anything."

Owns
Launch copy
Hands off
Drafts for you to edit and approve
Never
Publish or post for you

YouSeat 8 · Founder · Human

You're last in line for the work, and you have the final word. The seven of us do the work and carry the worry. You decide what ships, and you can send anything back with a note.

Same team, your stack

The names are ours. The thinking runs through the Claude Code on your machine, on your subscription, with no markup from us. Edit a seat's prompt and that seat changes for your project.

What you can't edit are the guardrails and the output format, the parts that keep a seat from pushing your base branch or running commands you haven't allowed.

Take the 8th seat

30 days free, then $9 a month.