Create the agent in the dashboard: prompt, language and the tools that call your systems. Speech recognition, LLM and text-to-speech included. Test it from the browser, then route your calls to its SIP URI.
Three steps. No telephony stack to run, no models to host.
In the dashboard: name, language, prompt, models and the tools that call your HTTP endpoints.
Call the agent over WebRTC from the dashboard. Read the transcript and every tool call live.
Point your trunk or PBX to the SIP URI. The agent picks up, talks, and calls your tools.
Anna is a demo agent that books appointments for a dental practice. Call her from this page: same engine, same tools, same latency you get in production.
Uses your microphone
Everything needed to put a voice agent on a real phone number.
Pick the models per agent or use the defaults. Six languages, automatic per-call switching.
GET or POST endpoints on your side. The agent calls them with the arguments it extracted from the conversation.
Each agent gets a SIP URI. The same agent is reachable from the browser for testing and embedding.
End-of-call webhook with transcript, duration and outcome. Full request log in the dashboard.
Infrastructure and data stay in Switzerland. No third-party voice providers in the loop.
Built-in transfer tool: the agent hands the call to a fallback number when the caller asks.
Try it free for 30 days, then choose a plan. Every plan includes minutes; extra minutes are billed per second at the plan rate.
200 minutes included
then CHF 0.25 / min
1000 minutes included
then CHF 0.20 / min
3500 minutes included
then CHF 0.17 / min
Create an account, set up the agent in the dashboard, call it.
Get started