Bizzit
An AI-powered business memory tool. Tags: AI, Mobile App, Solo Build.
The Problem
Every time I needed a plumber, a notary, or a restaurant a friend mentioned, I found myself digging through WhatsApp threads, old emails, screenshots, Notes, saved Instagram posts, and half-remembered conversations.
We do not save useful business info when we discover it because entering it properly is tedious. So we end up starting from scratch, asking the same questions again, or settling for whatever Google serves up.
How I Built It
I wanted capture to feel effortless. Snap a photo of a storefront, and the app uses OCR and geolocation to figure out what business it is. Or type something vague like 'ventura plumber brussels' and an AI agent handles the rest, pulling in addresses, phone numbers, websites, and generating a structured contact card.
The backend runs on Google Cloud with Cloud Run and Firebase, with Weaviate as the vector database for semantic search. The AI layer uses a router and orchestrator agent pattern, experimenting with both Mistral and OpenAI to balance quality and cost.
What I Am Learning
The photo-to-contact feature genuinely delights people, but delight on first use does not guarantee retention.
User interviews revealed that people are surrounded by others who have information they will eventually need, but they do not know who has it or when they will need it. The desire to share exists, but there is no good moment to share to.
This is an invisible problem. People experience the frustration privately but do not search for solutions online, which makes typical SEO and GEO acquisition tricky.
Roadmap
Right now I am focused on user acquisition, testing Reddit and Indie Hackers as channels, and exploring Answer Engine Optimization for AI discovery.
Open questions: Is personal organization the right wedge, or should I pivot toward the social and sharing insight? What is the minimum number of saved contacts before search feels valuable?