Get started

Step 1

Create your Breakpoint Engine account

Sign in or create an account. Your account is what your AI authenticates against when it connects.

Step 2

Add credits

Choose the pack that matches how much experimentation your raise needs — Explore ($29) to understand your structure, Round ($99) for an active raise, or Raise ($249) to pressure-test complex or extended financings. Details on pricing.

Right now you can skip this step. The first 100 founders to create an account get a free Round pack — it is already in your balance when you sign in, no card needed. The free units expire 31 December 2026, so they are meant for a raise you are running now.

Step 3

Connect your AI

Every client connects to the same hosted service with the same account — pick the one you already use. Claude and ChatGPT are the easiest paths; the editors are advanced workflows for technical founders.

• Claude step-by-step — add Breakpoint Engine as a connector, complete sign-in, and ask your question in chat.
• ChatGPT step-by-step — add the connector where your OpenAI plan supports custom MCP apps, then enable it in chat.
• Cursor step-by-step — paste into .cursor/mcp.json, reload, click Connect:

  {
    "mcpServers": {
      "breakpointengine": {
        "url": "https://cloud.breakpointengine.com/mcp"
      }
    }
  }

• VS Code step-by-step — .vscode/mcp.json with "type": "http" and the same URL (VS Code 1.101+, Copilot Agent mode).
• Claude Code step-by-step:

  claude mcp add --transport http --scope user breakpointengine https://cloud.breakpointengine.com/mcp

  Then /mcp to finish sign-in.

Step 4

Ask your financing question

Try a first prompt like:

I'm raising our Series A. Here is our current capitalization and the proposed term sheet. Use Breakpoint Engine to explain what these terms mean for founder common and show me what changes if we negotiate the option pool from 15% to 10%.

Your AI reasons → Breakpoint Engine calculates → your AI explains the result. Your AI previews what a calculation will cost before running it, and every answer comes from a deterministic solve — not language-model arithmetic. Need help shaping your capitalization inputs? See preparing capitalization inputs.