Personal Finance Tracker

A personal finance tool that connects to real bank and credit card accounts, delivers AI-powered financial briefings, and supports conversational follow-ups about your spending. All data stays local — nothing leaves your machine.

Every number comes from deterministic computation against your actual transaction data. The AI adds narrative and insight — spending patterns, category breakdowns, anomaly highlights — but never does the arithmetic. You get the story of your money, backed by numbers you can trust.

How it works

• Connects programmatically to your financial institutions
• Syncs transactions and balances to a local database on your machine
• Generates AI financial briefings — weekly summaries, category trends, spending alerts
• Supports conversational follow-ups: ask questions about your own spending data

Technical approach

Built on a local-first architecture — all financial data lives in a local DuckDB database, never on a remote server. The system uses a multi-provider LLM architecture, choosing the best model for each task: summarization, categorization, conversational Q&A.

The separation between deterministic computation and AI narrative is a core design principle. Aggregations, totals, and category breakdowns are computed directly against the database. The AI layer wraps those numbers in readable context — it never invents a figure.

Current stage

Personal Tool — the founder uses this tool weekly for real budget decisions. The discovery phase (connectivity, data trust, refresh cycles) is complete. Current focus is proving sustained weekly usage as a genuine personal finance tool, not just a technical demo.

What's next

• Prove 4+ weeks of sustained weekly usage as a primary budget tool
• Decide on distribution path: open-source packaging, multi-user support, or expanded scope
• Richer AI briefing formats — trend visualization, spending forecasts
• Broader institutional coverage and refresh reliability