AI Product
Several AI models analyze the same high-stakes decision independently. The strongest thinking is merged into one Unified Brief you can defend.
Decision ThinkTank is for a decision with real stakes — stay or migrate, hire or restructure, build or buy — where one chatbot thread is not enough. You describe the situation. Several models read the same brief alone, so no single model’s framing quietly steers the outcome.
Every analysis uses the same three lenses: Risk, Reversibility, and Stakeholders. Models can ask follow-up questions before they finalize, and they can look things up. When you run more than one model, the strongest ideas are merged into a Unified Brief. Authorship is hidden while that brief is written, then attributed afterward, so credit follows the idea.
This started as Decision Copilot. The product is now Decision ThinkTank, live at decisionthinktank.com.
Model Studies is the research program behind the product. Frontier models run the same decision scenarios and are blind-coded on a fixed rubric. The public results cover 3 studies, 5 cases, 4 models, 60 single-model decision briefs, and 120 multi-model unified briefs.
The studies change one condition at a time: how the user frames the story, whether provider names are visible when briefs are merged, and whether a recommendation stays stable when the same scenario is run again.
One finding changed the product. When a synthesizer could see who wrote each draft, credit moved with the brand, not only with the strength of the idea. Decision ThinkTank now writes the Unified Brief with authorship hidden by default.