WattAI is a small, open-source decision model that turns unstructured text into typed decisions, without generating text.

Send WattAI a piece of text and a list of options, and it tells you which option fits best, along with a calibrated probability for every option.

That makes it useful for the part of an AI pipeline where you don't need a chatbot, you need a reliable decision.

Classify a support ticket. Detect intent. Route a document. Score urgency. Decide whether something is a refund request. Categorise an agent trace. Build structured signals from text.

Instead of asking an LLM to generate an answer and then parsing it, you define the answers you actually want and let WattAI score them directly.

No generation. No JSON parsing. No prompt gymnastics. Just decisions.

Categorize runs locally, so you can use it offline and keep your data on your machine. The recommended model is a 726 MB F16 build, with smaller quantized builds available for constrained hardware and edge devices.

It also comes with a hosted API if you don't want to run the model yourself. The API is free during the v0.5 preview, with no account or API key required for the anonymous tier.

Give WattAI text. Give it the choices. Get a decision and the probability distribution behind it.