CHOOSEConsumer & creative
AI-Powered Puzzle Game Bot
Michaël Ménard@mikemenard_com
2:10A bot plays a puzzle game by querying an AI model for tap probabilities before each move, utilizing live API calls for decision-making.
SignalMy code lists the legal taps, Jev returns a probability for each, code taps the best one. 15 puzzles, 190 calls, ~240 ms per call, total cost $0.0298. (Most of that was the last hard puzzle) Why so fast and cheap?
Pattern↑ Jev AI decision engine
I built a bot that plays a puzzle game by asking an AI before every tap. Each tap you see is a live API call.
This is Jev, @typesafeai model. No prompt, no generated text. My code lists the legal taps, Jev returns a probability for each, code taps the best one.
15 puzzles, 190 calls, ~240 ms per call, total cost $0.0298. (Most of that was the last hard puzzle)
Why so fast and cheap? A normal LLM answers by writing. It thinks out loud, one word at a time, and every word costs time and money. Jev never writes. You give it the situation and a fixed set of options, and it reads once and hands back a number for each option: how likely it is to be the right one. Reading is fast. Choosing is instant. You only pay for what it reads.
That changes what you build. A judgment that costs a tenth of a second and a hundredth of a cent can sit inside a game loop, a search, a form, anything. Code keeps doing what code is good at: rules, math, proofs. The model only answers the questions that need judgment.