Skip to main content
Egma tests and monitors Pipecat bots, whether they run on Pipecat Cloud, on your own servers, or on your computer while you build them. Add the Egma SDK to your bot. It tells Egma which tools your bot has, answers the tools a test mocks, and sends your bot’s own record of each conversation to Egma.

​
How Egma works with Pipecat bots

​
For simulation

For each simulation, Egma sends one start request, the same request a Pipecat client sends to start a bot. Your starter creates a Daily room, starts your bot in it, and answers with the room’s address. Egma’s persona joins that room and talks to your bot. You can connect a Pipecat bot in three ways:
  1. Pipecat Cloud - give Egma your agent name and a public API key. Pipecat Cloud creates the room and starts the bot.
  2. Self-hosted - give Egma the public HTTPS URL of your own bot starter and the auth headers it checks.
  3. This machine - run egma agent dev beside your bot on your computer. It opens a tunnel to Pipecat’s development runner and fills in a self-hosted connection for you. No deploy is needed.
You can test your bot in two modalities:
  1. Voice - runs a full voice conversation through your bot’s speech-to-text, LLM, and text-to-speech.
  2. Chat - the persona sends text through RTVI and reads your bot’s text replies. Use it to test your bot’s reasoning and tools. It needs no code beyond the SDK line.
Every Pipecat simulation uses the Daily transport.

​
For monitoring

Add monitor(worker, runner_args) to your bot. It sends each production conversation to Egma, where you can read it and run graders on it.

​
Guides