Implement the AI Agent in your own apps via Server-Sent Events (SSE).
This endpoint allows you to send a prompt to the Chatmu Agent and receive a real-time streamed response. This is the exact same engine that powers Chatmu.io.
POST https://chatmu.io/api/chatmu-streaming
Content-Type: application/json
Authorization: Bearer YOUR_API_KEY
{
"message": "What is the demographic profile of Billie Eilish?",
"chatId": "uuid-for-memory-optional",
"tenantId": "your-workspace-id"
}
The response is a standard SSE (Server-Sent Events) stream. You will receive markdown text chunks representing the AI's answer.
However, the Chatmu Agent does more than text. When it executes analytical tasks, it returns internal JSON blocks so your frontend can render UI Widgets (like bar charts, usage reports, or the planner's tasks). These blocks look like this in the stream:
[CHATMU_AGENT]{"agent": "usage", "type": "usage_report", "payload": {"cost": x}}[/CHATMU_AGENT]How to handle it: Regex match [CHATMU_AGENT](.*?)[/CHATMU_AGENT] to extract the JSON. Depending on the type (e.g., usage_report), you can hide that raw string from the user and instead build a native UI component with the payload data!