Is it ready to ship?

• Developers holding a release candidate that “works on my machine”
• Product teams shipping a feature nobody has tried to break yet
• Agencies and freelancers about to hand finished work to a client
• Anyone whose work is technically done — but who can’t tell if it’s ready

1. 1

   Opening positions

   Each member reviews the work you paste or attach and stakes a position — ship or hold — grounded in specifics: the exact flow, edge case, security gap or unhandled state that decides it, never generic checklist talk.

2. 2

   Rebuttals & cross-examination

   The pragmatist’s “ship it, iterate later” is forced to answer the pre-mortem’s specific failure scenarios; the security engineer and QA auditor contest which findings are real blockers versus polish. Every member must add something new each round, not repeat.

3. 3

   Moderator synthesis

   The impartial chair issues the readiness verdict: GO, GO-WITH-FIXES or NO-GO — stated first, then the evidence. Blockers come ranked P0–P3, each with the concrete fix, alongside what is genuinely ready and shouldn’t be touched.

4. 4

   Final QA & sign-off

   The Final QA audit — a separate model outside the debate — checks the verdict for invented issues, overconfidence and missed caveats, and itemises what you must verify in the real system before relying on the call. Flags are shown, never hidden.

5. 5

   Council roll call

   You see each member’s final stance — agree, concern or dissent — so a GO that carried three members with two concerns reads exactly as that.

The shape of the verdict — illustrative structure, not a sample result. The content of each part comes from your brief and the council’s actual debate.

Each seat runs on a different frontier model — GPT, Claude, Gemini, Grok — so the debate is a genuine cross-model argument, not one mind in five costumes. Swap any seat from Decidi’s 86 personas.