Truth needs friction.
Most chat assistants are designed to agree with you. DebateGPT is designed to push back — six voices, one crux, one Decision Brief.
- Speakers
- Six parallel voices
- End state
- Decision Brief
- Privacy
- No account · local by default
- Mara Power Attorney
- Quinn Business Titan
- Reid Screen Icon
- Sloan Tech Magnate
- Pax Contrarian
- Atlas Heavyweight
The premise
Single-voice AI is great at sounding confident. It is bad at admitting when a question has no clean answer. A panel of six voices is closer to how decisions actually get made — by people with different priors, biases, and incentives, who pressure each other until one assumption breaks. The Assumption X-Ray and Decision Brief make that break point explicit: fragile assumption, mind-change trigger, reversible risk, irreversible risk, and next move.
What it is not
- It is not a fact-check engine. Truth Factor and Assumption X-Ray score and summarize the argument, not the world.
- It is not a chat companion. There is no memory across sessions and no pretend-friendliness.
- It is not a content farm. Six speakers, one crux, one Decision Brief, a clean exit.
Cost and privacy posture
DebateGPT works in a $0 local fallback mode for demos, testing, and visual QA. In live mode, it favors direct official providers and low-cost model routing instead of unreliable free aggregator quotas. Private session keeps snapshots out of browser history, and there is no account system or database.
Who it is for
Operators, lawyers, founders, policy people, journalists, and anyone who has to make a decision they cannot delegate. If you need a decision artifact rather than a reassuring single answer, this is the product.
Ask a question →