Open source · MIT licensed
Financial documents, read properly.
Upload a bank statement and get back categorized transactions, a cashflow picture, a risk score, and a policy decision. Through one API, three SDKs, a CLI, and an MCP server.
Quickstart
Create a submission, upload a statement, and read the analytics in about five minutes.
Concepts
Submissions, documents, transactions, policies — how the data model fits together.
API reference
Every endpoint, generated from the OpenAPI spec, with a request playground.
MCP server
Expose Moneyline to Claude and any MCP-compatible agent.
Build with an SDK
TypeScript
Type-safe client with auto-pagination and typed errors.
Python
Sync and async clients backed by httpx and pydantic.
Go
Idiomatic client with context support and functional options.
What happens to a document
1
Ingest
Upload a PDF, or pull statements directly from Plaid or a connected CRM.
2
Classify
Every page is typed — bank statement, tax return, P&L, driver’s licence — and split into logical documents.
3
Parse
Tables, balances, and transaction lines are extracted with per-field confidence.
4
Enrich
Transactions are categorized, merchant-matched, deduplicated, and checked for recurrence.
5
Decide
Run a scorecard, a risk score, or a full policy evaluation against the enriched data.
Moneyline is MIT licensed and can be run entirely on your own infrastructure.
See self-hosting.