Core decision
A language model reads each message but can only propose actions from a closed list. Code validates them against the case, and each insurer's rules, which are versioned files, decide where the case goes. Coverage decisions belong to a separate AI adjuster that shows its steps and the sources for each one.
How a message travels
One turn: reporting a crash
- turn
- input guard
Recognizes orders to ignore the rules and off-topic messages before any model is called.
- agent (Claude Haiku 5.5)
Understands the message and writes the reply. It can only propose actions from a closed list.
- case workflow (LangGraph)
Code validates every proposed action against the case, along a fixed workflow.
- insurer rules
Versioned files, one per insurer, each rule with its own test. The rules, not the model, decide where the case goes.
- policy search (pgvector)
Cites the exact clause. A verifier checks that it exists, belongs to that policy and is in force.
- AI adjuster
Takes the cases the rules do not close, decides and explains each step. The agent never decides coverage or payments.
- output guard
Checks every message, in both languages, before the person sees it.
Another turn: a question unrelated to insurance
- turn
- input guard
Answers with a fixed phrase from the catalog, in both languages, with no second model call.


In numbers and controls
- More than 4,000 automated tests with an 80% coverage floor.
- Insurers isolated with row-level security, and a test that fails if any table lacks it.
- Hard spending limits: prepaid balances, an application-level ceiling and daily limits per visitor.
How it fails
- Evaluation: 258 cases recorded in the latest full run (October 15, 2026). None failed.
- Chat latency, from the moment the server receives the message to the first word: p50 1.6 s and p95 2.3 s in Spanish (87 turns); p50 1.5 s and p95 2.2 s in English (85 turns).
- Voice latency: not measured.
- It is not connected to any real insurer. Voice works in the browser only, with a daily call limit.