Lucas Clemente
Business Problem:
Agencies and advanced users increasingly use AI assistants (Claude, Claude Code, ChatGPT, Cursor) to build, configure, and maintain customer workspaces. Today this creates two distinct blockers:
  • For everything the REST API covers (contacts, messages, custom fields, broadcasts), users must write custom scripts every time instead of asking the assistant directly.
  • For everything the API does not cover (Workflows, AI Agents, Snippets, channel configuration), the work is 100% manual in the UI — no AI assistance, auditing, or automation is possible at all.
  • For agencies managing dozens of client workspaces, this multiplies onboarding time and increases configuration errors.
Desired Outcome:
An official respond.io MCP (Model Context Protocol) server that allows any compatible AI assistant to connect via a standard connector and execute actions in natural language without writing code.
  • Contact management: Search, update, tag, and move contacts across lifecycle stages via natural language.
  • Messaging and broadcasts: Send broadcasts to tagged segments, read and summarize unanswered conversations.
  • Snippets: Create, list, and detect duplicate or outdated snippets programmatically.
  • Audit and diagnosis (stretch goal): Read and inspect Workflow and AI Agent configurations to assist with setup, auditing, and cross-workspace comparison.
Minimum actions the MCP should expose:
  • Read / create / update contacts, custom fields, tags, and lifecycle stages
  • Send individual messages and broadcasts
  • Read conversation history and internal comments
  • Programmatically create, list, and edit Snippets
  • (Stretch goal) Read and inspect Workflow and AI Agent configuration for auditing and setup assistance