Launching today
Polymetis
Predict how customers react before you change a price
2 followers
Predict how customers react before you change a price
2 followers
Describe a pricing, fee, auction or incentive decision in plain English. Polymetis predicts how buyers, sellers, bidders or sales reps will actually respond, using a behavioural game-theory model fitted on about 120,000 real human decisions, and shows the textbook answer next to it. Compare options, find the best setting, drag what-if sliders, and seal a decision record you can score later. Free, no sign-up.



Launch Team

AWS Activate CreditsCredits that fuel your startup’s next milestone
Promoted
Hi Product Hunt! I'm Meet, and I built Polymetis.
Most pricing tools assume customers respond the way the textbook says. They don't. People reason a step or two ahead and choose noisily. Raise a commission and sellers leave faster than theory predicts; cut a fee and they keep more of the saving than you'd expect.
Polymetis is a behavioural game-theory model (quantal level-k) fitted on about 120,000 real human decisions from published experiments. You ask in plain English (“should restaurant commission be 18% or 24%?”). A language model only translates the question into the model's inputs; every number comes from the behavioural model, shown next to the textbook answer.
Free, no sign-up:
Compare designs: 18% vs 24% commission, sealed vs open auction, one big prize vs five small ones
Find the best setting of a lever, with the full response curve
Drag sliders in a what-if sandbox
Seal a decision record and later report what actually happened, so predictions get scored
Honest limits: the general model captures typical human reasoning, not your specific customers. Calibrating to your own data is where it gets precise, and that's what pilots are for.
The benchmark and code are open: github.com/mmehta4/human-strategic-play
I'd love blunt feedback, especially if you've ever set a take rate, run an auction or designed a sales incentive.