Developers & AI agents
Plug vertical intelligence into your stack: REST APIs are live on the API gateway. A Model Context Protocol (MCP) surface— the "Judgment API" your agents call as a tool—is on our public roadmap so the same trust layer works from OpenAI, Copilot, Claude, Google, or custom orchestrators.
Integrate today (REST)
Primary query: POST /ask
Send a user query with organization context; receive an answer, citations, retrieval trace, conversation id, and audit id. Ideal for assistants and internal tools that need grounded, auditable responses.
Paths are rooted at your deployed API gateway (not under /api). See the full list on the API reference page.
Cases, ingest & audit
Case ingest and processing, decisions, overrides, and audit export endpoints support insurance, healthcare, and logistics workflows with a defensible trail.
Browse documented routes →Roadmap: MCP & agent tools
We are aligning with the same open standards described in our platform vision: hosted MCP tool endpoints, discoverable schemas, and optional packaged skills for major agent platforms—without locking you to one vendor.
Judgment as a tool
Structured request/response schemas for expert-backed or policy-grounded decisions, callable from any MCP-aware agent.
Enterprise-safe
Auth and tenancy consistent with your existing AVI deployment; audit trail preserved for agent-originated calls.
Composed workflows
Agents retrieve from your knowledge layers, then escalate or record outcomes through the same APIs you use today.
Interested in MCP beta or design partner access? Contact us or request enterprise access.
Human-in-the-loop task queues and SLAs for expert review complement these APIs in pilot and enterprise programs.
Pilot program