Team Sessions
AI Team Sessions
Teammates, AI, and non-technical stakeholders all contribute in the same session. Product managers describe what they want in plain English. Designers weigh in. The plan gets built collaboratively — with an AI that knows your code, your database, and every decision you've made.
payments-api — team session
Ali, Priya, Tom + DevSpec
Priya
Stripe webhooks keep timing out on retries. Can we just stage a fix for the backoff and get it out today?
The retry loop is in lib/payments/webhooks.ts:42— fixed 2s delay, no jitter. You decided in May that webhooks verify out-of-band, so the fix is safe to stage. I've written the acceptance criteria; rush it to the top?
Done. Pushed as [devspec:214], deployed to staging, and the AI test walked all three criteria — passing, on video. Want the link?
— a session, start to shipped. The whole team was in the room.
AI chat is single-player everywhere else
Each teammate asks the same questions in a private window, gets a slightly different answer, and the real decision still happens in a meeting afterwards. The context that mattered lives in five separate chat histories nobody can search — and the AI cheerfully agrees with whoever it's talking to.
A product manager, a designer, and a developer are in the same DevSpec session. The PM describes the feature. The designer refines it. The developer asks one clarifying question. Autopilot ships it.
How it works
A session is a real conversation with your whole team in it — and an AI participant that has done its homework.
Start a session in any project
The AI joins with full context already loaded: your indexed codebase, your database schema, open action items, and the memory of every decision your team has recorded. Nobody pastes in background.
Starting a project session — the AI greets the team with project context already loadedIt already knows the project. The session starts mid-thought.
Bring the whole team
Invite teammates into the thread, @mention someone to pull them in, settle disagreements with a quick poll, or talk it through in voice mode. Conversations that start in Slack can become sessions with a single mention — thread captured, participants matched.
@mentioning a teammate and running a poll inside a live sessionDisagreement becomes a poll, not a meeting.
Decisions get captured as you go
While you talk, the AI records the decisions, conventions, and risks as durable memories, turns conclusions into action items with acceptance criteria, and drafts step-by-step plans you approve before anything runs. The conversation is the documentation.
The AI capturing a decision as a memory and creating an action item mid-conversationNothing useful evaporates when the call ends.
Built for how teams actually talk
Voice mode
Talk naturally about architecture and features. Live transcription and spoken responses make a session feel like a meeting with an AI team member who actually knows your code.
@Mentions
Pull a teammate into the thread when their call is needed. They're notified and land in the conversation with the full history — no recap required.
Polls
Split on approach? Run a quick poll in the session. The options, the votes, and the outcome stay attached to the decision they settled.
Plans with approval gates
For bigger work, the AI drafts a multi-step plan. You review, edit, and approve each step — judgment stays human, execution gets delegated.
Slack to session
Start in Slack, finish in DevSpec. One @mention captures the whole thread, matches the participants, and opens a session with the discussion intact.
Catch-up
Log back in and instantly know where things stand: what teammates did, what Autopilot shipped, what was decided. No trawling Slack threads or commit logs.
You were heads-down elsewhere for two days. You scan the catch-up summary and know exactly what needs your attention — in under a minute.
Convinced?
Set up your first project in under five minutes. Connect your repo, start a session, and see what your duck can do.
Start Your Project14-day free trial · 1,500 credits included