Build financial models with Claude. Scale beyond spreadsheets.
Describe business logic in plain English. Claude proposes Python through Bridge Town MCP tools. Bridge Town runs it in isolation. Review every answer in your chat.
Connect to Claude- Free to start
- No credit card
- Google login
Using Claude Code, Codex, Cursor, or another MCP client?
Full install docs →
Live in Claude · bridge town connected
Why is FY27 EBITDA $6.2M? Walk me through it.
Gross profit of $28.5M minus $22.3M of operating spend. The biggest lever is headcount — you reach 84 FTEs by December. Want me to branch a hiring-freeze scenario?
Yes, freeze hiring at 76 and show me the impact.
Branched, ran, and opened change #143 for your review — main is untouched until you approve.
Share the model with my CFO jane.doe@acme.com so she can review the freeze.
Done — Jane now has editor access. She can open the sheets, trace any number, and approve change #143 herself.
/ under the hood
This isn't just a chat.
Every answer Claude gives comes from a real financial model in your workspace — inputs, outputs, scenarios, and change requests you can open, share, check, and approve.
| $K | Jan | Feb | Mar | Apr | May | Jun | FY27 |
|---|---|---|---|---|---|---|---|
| Revenue | |||||||
| Net revenue | 3,420.0 | 3,495.0 | 3,572.0 | 3,651.0 | 3,731.0 | 3,813.0 | 43,800.0 |
| COGS | 1,197.0 | 1,223.2 | 1,250.2 | 1,277.9 | 1,305.9 | 1,334.6 | 15,330.0 |
| Gross profit | 2,223.0 | 2,271.8 | 2,321.8 | 2,373.2 | 2,425.1 | 2,478.5 | 28,470.0 |
| Operating expenses | |||||||
| Headcount | 1,215.0 | 1,222.0 | 1,229.0 | 1,237.0 | 1,244.0 | 1,252.0 | 14,905.0 |
| S&M + G&A | 601.0 | 604.0 | 607.0 | 611.0 | 614.0 | 618.0 | 7,361.0 |
| EBITDA | 407.0 | 445.8 | 485.8 | 525.2 | 567.1 | 608.5 | 6,204.0 |
01 · native sheets output sheets are written from the model; the same inputs drive the same outputs.
| driver | FY26 | FY27 |
|---|---|---|
| arr_growth | 24% | 22% |
| gross_margin | 64% | 65% |
| ftes_eop | 71 | 84 |
02 · input sheets each edit creates a version you can trace to a human.
03 · scenarios branches of one model — compare, merge, or discard.
backfill_attrition = trueftes_eop = 84ftes_eop = 7604 · version control Your AI client proposes; a human approves. Roll back anytime.
/ how it works
One click. Three minutes.
From nothing to a working model Claude can answer questions about — in about the time it takes to write the request.
Connect to Claude.
One click and one sign-in. The connection works across Claude on web, desktop, and mobile — no setup, no tokens to copy.
Describe the model.
Tell Claude what you need in plain English — "a 3-statement model for a SaaS company at $40M ARR." It proposes the model through Bridge Town, which stores and runs it and shows the results in seconds.
Review and approve.
Every change comes to you as a proposal with its impact on EBITDA and cash attached. Approve it, send it back, or roll back to any earlier version.
Your next model is one click away.
Free for your first workspace. Production teams scale to seats.
Connect to ClaudeUsing Claude Code, Codex, Cursor, or another MCP client? Full install docs →
Your provider, your agreement. Bridge Town does not run or proxy a language model. Your chosen AI provider processes the conversation under your agreement; Bridge Town receives only the MCP tool calls you authorize.
You decide who’s in, and for how long. Your chosen AI client connects to your workspace only after you approve it, sees only what your account can see, and can be disconnected in one click — instantly.
Locked up the way your bank does it. Encrypted everywhere it travels and everywhere it rests, with keys unique to your workspace.
Built to survive your IT review. Every access and every change is logged with who, what, and when — the documentation your auditors ask for already exists.