Resources
Bridge Town is a financial modeling platform built around the Model Context Protocol (MCP). AI agents connect to Bridge Town's MCP server to read, write, and run versioned Python finance models — the same way they interact with code repositories and developer tools. This page explains what MCP-native means in a financial modeling context and why it matters for FP&A teams.
Try Bridge Town FreeThe Model Context Protocol is an open standard for connecting AI agents to tools, data sources, and services. An MCP server exposes a set of structured tools — operations that an AI agent can call with typed inputs and receive typed outputs. The agent does not need custom code for each integration; it uses the protocol.
Bridge Town's primary interface is an MCP server. This means AI coding agents — including Claude and other MCP-compatible runtimes — can perform financial modeling operations through the same protocol they use for software engineering tasks.
Most financial planning tools treat AI as a feature layered on top of an existing product — a formula suggestion, a natural-language query, or an export to a chat interface. MCP-native means the opposite: the AI agent is the primary actor. It works with financial models the same way a software engineer works with a codebase, through a structured protocol with full read-write access and a version-controlled audit trail.
Analyst describes the model
An FP&A analyst types a planning request in plain English — for example, "build a headcount forecast that projects salary cost by department for the next 12 months."
AI agent authors the model through MCP
Bridge Town's AI agent uses the MCP server to write Python model code, connect input data, and configure the run. The agent has structured access to project state and can inspect existing assumptions before writing.
Model runs and outputs are versioned
The model runs inside Bridge Town's sandboxed compute environment. Outputs are captured and committed to the project history alongside the code that produced them.
Finance team reviews the output
The analyst reviews outputs in the Bridge Town interface, asks the agent to adjust assumptions, and publishes the final version. Every change is tracked in the project history.
MCP stands for Model Context Protocol, an open standard that lets AI agents interact with tools and data sources through a structured interface. An MCP-native financial modeling platform is one where AI agents can natively read model files, write Python logic, run models, and retrieve outputs through the protocol — rather than through a bolt-on integration. Bridge Town's primary interface is an MCP server, which means coding agents can work with financial models the same way they work with code repositories.
Traditional financial software AI integrations are add-ons layered on top of an existing tool. The AI can suggest formulas or generate summaries, but it cannot inspect model state, run the model, or commit changes. MCP gives an AI agent structured access to all of those operations. In Bridge Town, an AI agent can read the current model assumptions, modify specific cells or functions, trigger a model run, and inspect the output diff — all through the protocol without custom scripting.
Any coding agent that supports the Model Context Protocol can connect to Bridge Town's MCP server. This includes Claude, agents built with the Anthropic SDK, and other MCP-compatible runtimes. The MCP server exposes tools for reading project state, writing model logic, running models, and fetching outputs.
No. Finance analysts describe their planning logic in plain English. Bridge Town's AI agent writes the Python, connects the data, and runs the model. The MCP layer is the interface used by the agent, not by the analyst directly.
Because AI agents interact with Bridge Town through a structured protocol, every operation — reading assumptions, writing formulas, running models, committing changes — is logged in the project history. Finance teams can review what the agent did, what changed, and what the output was for any model version. This creates an auditable record that is not available when AI tools generate content outside a version-controlled environment.
Bridge Town is free to start. Describe a planning workflow in plain English and get a running Python model in your first session.
Start FreeRelated resources