Allow developers to connect any external AI agent to Ringee and make or receive phone calls. The agent can run anywhere — on a laptop, server, Claude, ChatGPT, Gemini, a local LLM, or a custom application — while Ringee provides the telephony layer. Agents should be able to connect through API, WebSocket, SDK, CLI, or MCP and control calls in real time, including: Start and end calls Receive live transcripts and call events Send text responses that Ringee converts to speech Support interruption/barge-in Send DTMF and transfer calls Optionally stream raw audio for custom STT/TTS stacks The goal is simple: Bring your own agent. Ringee gives it a phone.