Genie
Documentation

Genie Chat

One agent, two front doors. Talk to Genie in the browser, or call the OpenAI-compatible API — same agent, same skills, same memory either way.

The web chat

Open Genie Chatand start a conversation — nothing to install. The agent has the same skills and the same memory it has over the API, so a thread you start in the browser carries the context you'd expect.

The API

The same agent answers over the OpenAI-compatible endpoint, so any OpenAI SDK works unchanged — just change the base URL:

bash
curl https://api.genie.tech/v1/chat/completions \
  -H "Authorization: Bearer sk-genie-{your-key}" \
  -H "Content-Type: application/json" \
  -d '{"model":"genie","messages":[{"role":"user","content":"what broke the deploy last night?"}]}'

Full request/response shape, streaming, and async are documented under Chat completions. Get a key from /settings/api-keys — see Authentication.

One agent, not two

  • Same skills and tools whether invoked from the browser or the API.
  • Same routing — requests run on the persona-routed fleet, on local open-weights models.
  • Load shared context into any conversation with Genie Context (genie context get).