I leveraged AI to do my taxes, and for the first time in 12 years of owning LLCs, this year's preparation was the fastest, cleanest, and most professional I have ever had. Equipped with GitHub and Codex/Claude, I built a personal financial aggregator dashboard in just a few hours that pulled together bank statements, credit-card statements, business-versus-personal expense categorization, and auto-rules that pushed my federal tax preparation toward autopilot.
Most people treat taxes like a seasonal emergency. I wanted the opposite: a system. This project was about using GitHub and Codex to build the infrastructure before filing season became painful: aggregate financial data, categorize spending automatically, preserve the logic in version control, and arrive at federal tax prep with cleaner books, sharper visibility, and far less friction.
A personal finance dashboard and supporting rule engine that organizes statements across accounts, tracks spending patterns, and keeps business versus personal expense separation visible enough to support real tax preparation rather than just casual analytics.
The biggest win was not novelty. It was professionalism. I have never prepared my taxes this quickly or this cleanly, and the system now improves over time because categorization logic and review decisions can live in the repo instead of only in my memory.
The sections below cover the original tax-prep pain, the finance system itself, and why GitHub plus Codex changed the quality of the workflow.
I did not build a robot to file my return blindly. I built a system that made the preparation legible, structured, and far easier to review. That distinction matters. The agentic part handled organization and operational throughput; the serious part was arriving at tax prep with cleaner books and better judgment.
