Strot - Compile LLM intent once, run decisions locally forever
Show HN: Strot – Compile an LLM's intent once, run decisions locally forever
Strot is an open decision compiler that lets you define your intent once with an LLM, then run classification, scoring, and routing locally on your own hardware at zero marginal cost. The LLM is used only at compile time to synthesize a frozen, type-safe Python artifact; production then arbitrates every record with a small local decision model in about 60 ms, producing JSON with probabilities and confidence. Low-confidence records abstain instead of guessing, and fitted guarantees keep dispatched errors bounded. Benchmarks show 92.5% routing accuracy versus 74.8% naive, with zero LLM calls per record. Install via pip install strot-ai.
Compile your intent once with an LLM. Then decide on every record forever — classification, scoring, routing — as type-safe JSON with probabilities and confidence, on your own hardware, at zero marginal cost.