Building Trip Threads: The NLP Journey From Regex to LLMs
How I evaluated four different approaches to natural language parsing for Trip Threads and what I learned about building AI-powered products.
A personal finance companion that provides deep insights into spending habits beyond simple totals.
Q1 2026
planned
A personal finance companion that provides deep insights into spending habits beyond simple totals.
Features CSV upload support for bank exports, invoice parsing for line-item granularity, and AI-powered categorization.
Includes WhatsApp/SMS nudges, weekly digests, and category-based spending goals.
Built with Supabase backend for auth, storage, and PostgreSQL database with row-level security.
Existing finance apps only show totals, not what you actually spend money on
Wanted deeper insights into grocery spending and subcategories (fruit, dairy, snacks)
Build a chat-first spending coach with granular, actionable insights
Create a privacy-first alternative that works with either CSV uploads or if users choose then we can include bank APIs
Blog posts about this project and related topics
How I evaluated four different approaches to natural language parsing for Trip Threads and what I learned about building AI-powered products.
Notes from building Trip Threads — a personal experiment in making travel planning and coordination simpler, smarter, and a bit more human.
How Claude Code on the web turned my dog walks into productive coding sessions.
How I rebuilt my workflow in 2025 | from chasing speed to building with direction.