Origin
FoppyAI grew out of Fopymes, the prototype my teammate and I built for the Hatary Shunko Fintech Innovation hackathon, where it placed 3rd. After the event, the two of us kept building on that prototype with more time and care, turning the demo into a more complete personal-finance application.
Problem
Logging expenses by hand is tedious enough that people stop doing it. We wanted a Spanish-language voice interface so a user could capture a transaction, savings goal, or budget in a single spoken sentence.
Architecture
Audio is recorded in the Next.js client and sent to a Hono backend running on Bun. OpenAI Whisper transcribes the Spanish speech; GPT-4o-mini classifies intent (transaction, goal, or budget) and extracts structured fields. A small set of agent services normalizes and validates the extracted data before it returns to the UI for confirmation. The backend follows a hexagonal (ports/adapters) layout with Drizzle ORM over PostgreSQL.
My role
I was the dominant contributor — roughly 73% of backend and 68% of frontend commits. I built the voice module (ports/adapters, the multi-agent orchestrator, the Whisper and GPT adapters, the validation layer) and the frontend audio-capture hook and voice UI.
Scope: a focused hackathon prototype — Spanish, voice-to-intent routing over three flows, backend-tested — built as applied AI integration in a fintech context.
