Capture leads on WhatsApp and save them to Sheets
An n8n WhatsApp agent asks leads for their name and email and automatically saves them to Google Sheets, shown in a demo.
A custom MCP server exposes list, schedule and cancel appointment tools so an ElevenLabs voice agent can handle customer calls against the company CRM.
A custom MCP server exposes list, schedule and cancel appointment tools so an ElevenLabs voice agent can handle customer calls against the company CRM.
The commenter's team runs a custom MCP server exposing tools such as list_appointments, schedule_appointment and cancel_appointment against their CRM. An ElevenLabs voice agent, which handles speech-to-text, the LLM and text-to-speech, connects to it over a Twilio phone line. A system prompt sets tone, guardrails and when to call each tool.
The source is a short explanatory comment and reports no results, numbers or shortcomings. It describes the architecture only.
The LLM runs on infrastructure you don't control, so your own hosted MCP server is the only bridge to your CRM; the voice transport (phone, app, browser websocket) is interchangeable.
Build an MCP server in [language] that exposes list_appointments, schedule_appointment, and cancel_appointment tools on top of [CRM or API name]. Make it usable by a voice agent such as ElevenLabs, and include tests with fake data. Stop when the tools pass the tests and give me a short guide for connecting it, with cancellations requiring confirmation.
An n8n WhatsApp agent asks leads for their name and email and automatically saves them to Google Sheets, shown in a demo.
An n8n agent handles customer WhatsApp messages, searches info, checks the calendar, logs to Sheets or Notion, and replies by WhatsApp and email.
A Lindy agent researches YouTube ideas and scripts, and another emails a brief on discovery-call contacts 10 minutes before meetings; complex content stayed on Make.