MicHelper

B2B SaaS platform for AI analysis of conversations
of salespeople in physical retail.

How it works:
— The salesperson wears a microphone during their shift
— The mobile app records conversations
— VAD filters silence (reducing API costs by 69%)
— Groq Whisper transcribes audio
— AI analyzes the quality of conversations and identifies issues
— The manager receives an automatic report every morning

Technical stack:
Node.js, React, TypeScript, PostgreSQL, Redis,
Groq Whisper v3, Claude/Qwen AI, Capacitor
(Android APK), REST API, WebSockets

288,000+ lines of production code.
Developed entirely independently in 1 year.
Active pilot in 6 stores.

#nodejs #react #saas #ai #javascript