Documentation
Build your first agent
From an empty prompt to a live, observable voice agent. Here is the path, end to end.
Quickstart
Create an agent
Start from a prompt for a single-purpose agent, or build a visual workflow when the conversation has branches, tools, and handoffs.
Choose providers
Pick the language model, speech-to-text, and voice for the agent. Use the platform defaults or bring your own provider keys.
Test in the playground
Talk to the agent live in the browser. Watch the transcript, the routing decisions, per-step cost, and latency as it happens.
Connect a channel
Ship the same agent to your app, the web, a phone number, or your backend over the API and SDK, no rewrites.
Observe in production
Every session is recorded as a full transcript with cost and timing, so you always know what happened and why.
Core concepts
Agents and prompts
Workflows
Observability
Knowledge base
Tools and integrations
Channels
Call the API
The REST API lives under a versioned base URL. Create an API key in the dashboard and send it in the X-API-Key header, then list your agents to confirm the connection:
curl https://api.glytos.com/api/v1/workflows \
-H "X-API-Key: $GLYTOS_API_KEY"A successful request returns a JSON array of your agents. Every resource you see in the dashboard is available the same way.
Embed a voice agent
To run an agent in the browser, mint a short-lived, workflow-scoped token on your server, then connect with the browser SDK. Your API key never leaves your backend.
curl -X POST https://api.glytos.com/api/v1/calls/web-token \
-H "X-API-Key: $GLYTOS_API_KEY" \
-H "Content-Type: application/json" \
-d '{"workflow_uuid": "..."}'import { GlytosWebCall } from '@glytos/web';
// token + ws_url come from the web-token call above
const call = new GlytosWebCall({ token, wsUrl });
await call.start();