The story behind the name
Meet Wilson.
Named after Frank J. Wilson — the forensic accountant who followed the money to convict Al Capone. The original "follow the money" guy.
The man who got Capone
In the 1930s, everyone knew Al Capone was guilty. Nobody could prove it. The FBI tried wiretaps. Local cops tried raids. Nothing stuck.
Then the IRS sent Frank J. Wilson. A quiet, meticulous forensic accountant. While others chased gunfire, Wilson chased ledgers. He spent years tracing cash through shell companies, coded bookkeeping, and laundered receipts.
Wilson proved that Capone owed $215,080 in taxes on unreported income. Capone went to Alcatraz. The conviction came not from bullets, but from spreadsheets.
That's the philosophy behind Open Accountant. The answers are always in the numbers. You just need the right tool to find them.
What Open Accountant does
Every finance app stops at charts. Open Accountant tells you what to do about it.
Privacy first
Your bank data never touches a server. Ollama runs AI on your machine. SQLite stores everything locally. Zero bytes uploaded.
Actionable intelligence
Not pie charts. Not dashboards. Concrete recommendations: cancel this subscription, switch this provider, you're overpaying here.
Open source
MIT licensed. Audit every line. No black boxes. If you don't trust the code, read it. If you can improve it, submit a PR.
Terminal-first
The Wilson CLI
A command-line AI agent that imports your bank transactions, categorizes them with local AI, and surfaces actionable intelligence. Built with Bun, powered by Ollama, stored in SQLite.
→ Import from Chase, Amex, BofA, OFX, QIF
→ AI categorization with 120+ Plaid-aligned categories
→ Skills: subscription audit, tax prep, spending review
→ Agent loop architecture — iterative, tool-calling AI
→ 9 LLM providers: Ollama, Claude, GPT, and more
$ wilson import ~/Downloads/chase-march.csv
> Detected format: Chase Credit Card
> Parsed 147 transactions (Feb 1 – Feb 28)
> Dedup: 0 duplicates skipped
> Imported 147 transactions.
$ wilson categorize --auto
> Categorizing via Ollama (llama3.2)...
> 147/147 categorized. 94% confidence.
$
Follow the money.
Install in 2 minutes. Import a CSV. See what your finances are actually telling you.