Review Distiller
An AI SaaS to analyze & monitor customer reviews
- Role
- Founder & full-stack engineer
- Status
- Live · no longer developed
Conceived and coded entirely solo - system design, frontend, backend, infrastructure, payment - in full-stack TypeScript. It analyzes and monitors customer reviews from Google Maps, Play Store, App Store and more, and uses LLMs to summarize, score sentiment and even draft responses. My real goal: experience first-hand what it takes to build, run and market a complex SaaS.
01The idea
I originally wanted to build an iOS app, and while researching competitors I slogged through endless app reviews. So I scraped them and fed them into GPT for a summary - and realized this was useful far beyond app stores.
Any company bringing change into the real world - like the computer-vision solutions in my day job - should be monitoring customer feedback. Review Distiller was born.
02What I learned
An incredible amount - system design, GCP, GitHub Actions CI/CD, Terraform, deeper TypeScript, MUI, Fastify, Prisma, Postgres, and endless debugging.
But the biggest lesson wasn't technical: a product alone isn't enough. Distribution, product-market fit and founder-market fit aren't buzzwords - they decide everything. Even though the go-to-market didn't work out, it was absolutely worth it.
Features
Features
- AI summaries & sentiment analysis of large review sets
- Multi-source: Google Maps, Play Store, App Store & more
- AI-drafted responses to individual reviews
- Topic-level satisfaction & metrics over time
- Theming, dark mode, EN / DE
Stack & Infra
- Full Stack TypeScript
- React + MUI
- Node.js + Fastify
- PostgreSQL + Prisma
- GCP (Cloud Run, Pub/Sub, Functions)
- Terraform, GitHub CI/CD, Firebase
- Paddle payment
- Extremely cost-optimized yet scalable system design
AI & concept
- Different models for cost optimization
- Function calling & tool use
- Prompt Engineering
- Context Management
- Pricing model: Review Points, a virtual currency