Overview
MetMe helps professionals keep track of the people they meet. Users record a voice note after a meeting, and the app turns it into a summary and key points, then makes it searchable in plain language.
The Problem
Traditional CRMs need a lot of manual typing, and finding a contact later means remembering exact names or fields. Users wanted to capture notes fast and find people the way they remember them.
Technical Solution
System Architecture
- •Next.js with TypeScript
- •Supabase (PostgreSQL) for data and authentication
- •OpenAI Whisper for voice-to-text
- •OpenAI GPT for summaries and key points, checked against client-defined bounds
- •pgvector for natural-language semantic search
- •Stripe subscription packages with admin dashboards for users, packages and pricing
Frontend
Next.js, TypeScript
Database
Supabase (PostgreSQL)
AI
OpenAI GPT, Whisper
Search
pgvector semantic search
Results & Impact
8 months
team build at InovativeX
Voice to notes
recordings become summaries and key points
Stripe
subscription packages with admin dashboards
Lessons Learned
Bound the AI output
Checking summaries against client-defined limits kept the content consistent and professional.
Voice UX needs care
Recording, upload and error states mattered as much as the model.
Search by meaning
Semantic search let users find contacts the way they remember them, not by exact fields.