IntégrationsMay 15, 2026·6 min read

Your proposals, directly inside your AI — MyPitchFlow launches its MCP connector

MyPitchFlow now connects natively to Claude.ai, Cursor and Windsurf via the MCP open standard. Your proposals, knowledge base and RFP answers are available directly in your AI — without copy-paste, without switching tabs.

Key takeaways

What it is: MyPitchFlow is an RFP response and proposal generation platform that now supports MCP (Model Context Protocol), the open standard for connecting business tools to AI assistants
Compatible with: Claude.ai, Claude Desktop, Cursor, Windsurf, VS Code GitHub Copilot — and any MCP-compliant client
5 tools exposed: `search_knowledge_base`, `list_proposals`, `get_proposal`, `get_proposal_versions`, `search_rfp_answers`
Setup: Settings → Integrations → Generate URL → Paste into Claude.ai. 2 minutes, no installation
Available from: Complet plan · Workspace admin only

Two worlds that didn't talk to each other

MyPitchFlow already has an embedded AI agent: proposal generation, section-by-section refinement, Ctrl+Click targeting, automated RFP answers. AI has been in the product from day one.

But there was an invisible barrier: when you worked in Claude.ai, Cursor or Windsurf, you were cut off from your data. Your AI assistant knew nothing about your live proposals, your knowledge base, your past RFP answers. Two parallel universes, no bridge.

The MCP connector is that bridge. MyPitchFlow is now a data source that any MCP-compatible AI client can query in real time.

What is MCP and why does it matter?

MCP (Model Context Protocol) is an open standard published by Anthropic in November 2024 and adopted within months by the major AI clients on the market. Its purpose: allow any business tool to expose its data and functions to an AI assistant in a standardised, secure way — regardless of which AI is doing the asking.

MyPitchFlow implements MCP as a hosted remote server (Streamable HTTP transport, JSON-RPC 2.0). The connector URL is unique to each workspace and can be revoked at any time.

Compatible AI clients:

| Client | Connection method | |---|---| | Claude.ai | Settings → Connectors → Add custom connector | | Claude Desktop | JSON config file | | Cursor | `.cursor/mcp.json` | | Windsurf | MCP settings | | VS Code GitHub Copilot | MCP extension | | Continue.dev | `config.json` |

Not compatible: ChatGPT (OpenAI uses a proprietary GPT Actions system).

Respond to RFPs in minutes with AI

MyPitchFlow generates your proposals from your own knowledge base.

The 5 tools available in your AI

Once connected, your AI has access to 5 tools that query your MyPitchFlow workspace in real time:

`search_knowledge_base`— Semantic search across your uploaded documents: case studies, methodologies, pricing sheets, certifications. Grounds answers in your real content, not generic responses.

`list_proposals`— Returns your proposal pipeline with one entry per client (always the latest version), status, and creation date. Filter by status: draft, sent, won, lost.

`get_proposal`— Full content of a specific proposal by ID. Useful for review, coaching, or using an existing proposal as a starting point.

`get_proposal_versions`— Complete version history of a proposal family, from V1 to the current version, with status per version and creation date.

`search_rfp_answers`— Full-text search across all RFP questionnaire answers in your workspace. Find how a specific technical question was answered in a past bid.

Version history: your commercial memory, finally usable

Every MyPitchFlow proposal keeps its full history — V1, V2, V3... Until now this information lived in the interface, accessible but rarely exploited.

With the MCP connector, Claude can navigate that history using `get_proposal_versions`. Two concrete use cases:

Client dispute resolution— A client challenges a clause. In 10 seconds, Claude traces which version introduced it, in what context, and what existed before. End of ambiguity.

Contextual coaching— When you ask Claude to help improve a proposal, it understands the journey from V1: the concessions made, the positions held, how the response evolved. Its help is strategic, not just editorial.

Why the open standard is an advantage for you

We could have built a Claude-only proprietary integration. We chose MCP because it is the right architecture for the years ahead.

If your technical team works in Cursor, it works. If your sales team prefers Claude.ai, it works. Your MyPitchFlow data is not locked into one AI ecosystem.

This mirrors exactly what Notion expressed with its External Agent API, announced in May 2026: the future is not one application that does everything, it is an ecosystem of specialised agents that collaborate around open standards. MyPitchFlow takes its place in that ecosystem today — not as a roadmap item, but as a live connector you can set up in two minutes.

Related Comparisons

Notion as AI orchestration layer RFP knowledge base guide Why MyPitchFlow

Frequently Asked Questions

Everything you need to know about AI-generated proposals.

No. Claude queries MyPitchFlow directly via your unique key on every request. Nothing is stored on the AI side. Your data stays in your Supabase workspace — the connector is a read-only bridge.

Not currently. OpenAI uses a proprietary system (GPT Actions) that requires a separate integration. The MCP connector is compatible with Claude.ai, Cursor, Windsurf, VS Code GitHub Copilot, and any MCP-compliant client.

The workspace administrator only. The key is scoped to the workspace — all members' data is accessible through it, so admin control is required.

Yes — Settings → Integrations → Revoke. The key is immediately invalidated. Any AI client using it will receive a 401 error on the next request.

Ready to write better proposals, faster?

MyPitchFlow generates professional proposals in 2 minutes. See it in action.

Personalized 15-minute demo