Connect any API
Bring your own systems.
Define an action as an HTTP call (GET, POST, PUT, PATCH) to your endpoint, with headers, auth, and request/response templates. Map values in and out using simple variables, and pull exactly the fields you need.
- GET, POST, PUT, PATCH. Call your endpoints with full control over headers and body.
- Templated in and out. Map chat variables into the request, pull fields from the response.
- The AI can run them. Fire from a flow node, or let the agent invoke them during chat.
Action · lookupOrder
GET /orders/{id}
MethodGET
AuthBearer · from vault
Inputorder_id ← chat
Returnsstatus, eta, items
Ran in 240ms · logged with full response
Safe by default
Control what runs, and what it can reach.
Approval gates, a masked secrets vault, and a host allowlist keep automation safe in production.
Approval workflow
Require approval before sensitive actions run, with pending, approved, and rejected states.
Secrets vault
Store API keys and credentials masked and encrypted, never exposed in chat.
Host allowlist
Outbound calls are restricted to hosts you trust (SSRF protection).
Test before you ship
Try it without touching production.
Run a test execution to validate an action, and use Luran's built-in sandbox endpoints (order status, account lookup, appointment booking) to prototype integrations before wiring up the real thing.
- Chain lookups. Use the result of one call as the input to the next, to resolve multi-step requests.
- Sandbox endpoints. Prototype against built-in order, account, and booking endpoints.
- Full execution history. Every run is logged with status, response, timing, and errors, and you can retry failures.
Execution log
lookupOrder200240ms
checkAccount200180ms
bookSlot409120ms
createTicket201300ms