What is a decision model?
A decision model is not a chatbot. You don't get paragraphs of text — you give it a state (a ticket, an email, a message, a JSON document) plus one or more typed questions, and it answers each one in a single forward pass:
- Choice — picks one of your options, with a probability for each.
- Score — an expected value over your ordered levels.
- Yes / No — the probability that something is true.
Because nothing is generated, there is nothing to parse and no format to repair. Typical uses: routing tickets, classifying email, scoring urgency, gating an LLM call.
Powered by Laya
This tool runs on Laya, the open-source System-1 decision engine from
Convai Innovations — a ~421M-parameter, non-autoregressive model released under the
Apache-2.0 license. It answers typed questions in tens of milliseconds
and can be self-hosted. The backend speaks the Jev-compatible
POST /v1/systemone wire protocol.
Project: github.com/NandhaKishorM/laya
Honest limits
- Probabilities are model estimates, not guarantees. Check calibration on your own data before trusting confidence thresholds.
- Laya only picks among the options you listed — it can't invent answers or explain its reasoning in prose.
- It's not a chatbot, a lawyer, a doctor, or a financial adviser. Don't use it for financial, legal, or medical decisions.