Contable AI Pro

Personal finance tools ask you to do the work. You categorize, you label, you reconcile. For anyone with years of banking history and hundreds of monthly transactions, that friction is enough to make the whole thing pointless. I wanted to build something where the user does nothing except drag a file.

A web application that takes raw CSV bank exports and turns them into a structured, categorized financial dashboard in seconds. The core engine is Claude 3.5 Sonnet, which handles semantic categorization of transactions by learning from the user's own consumption patterns over time. Not rule-based sorting. Actual reasoning about what a transaction means in context.

The technical decisions I'm most proud of: implementing Anthropic's prompt caching to reduce token consumption by 90% by reusing historical user context across sessions, and using Zod for strict data validation to ensure AI responses conform to the application schema. That last one matters more than it sounds. An AI that returns malformed data in a financial context isn't just wrong, it's a liability.

Next.js for frontend and API. Claude 3.5 Sonnet for financial reasoning. Vercel Blob Storage for CSV handling. Private authentication middleware and full search engine exclusion for data privacy.

Key signals from the first production iteration.

4,500 rows of raw financial chaos to an organized dashboard in milliseconds. A zero-friction tool that understands your financial life rather than just reading it.