Skip to content

Releases: MomenMushtaha/ReviewLensAI

πŸš€ ReviewLens AI v1.0.0 β€” Initial Release

15 Mar 04:32

Choose a tag to compare

πŸŽ‰ ReviewLens AI v1.0.0 β€” Initial Release

ReviewLens AI is an AI-powered review intelligence portal that transforms raw Trustpilot reviews into actionable insights. Paste a URL or upload a CSV, and get sentiment analysis, theme clustering, rating trends, and a guardrailed Q&A chat β€” all in minutes.

🌐 Live Demo: v0-review-lens-ai.vercel.app
πŸ“‘ API Docs: reviewlens-api.onrender.com/docs


✨ Features

  • Trustpilot Scraper β€” __NEXT_DATA__ extraction with JSON-LD fallback, paginated multi-page scraping, and CSV upload support
  • Smart Deduplication β€” SHA-256 hash of (source_url + body[:200]), scoped per project
  • Sentiment Analysis β€” VADER compound score with star-rating override for extreme ratings (1–2β˜… β†’ negative, 4–5β˜… β†’ positive)
  • Theme Discovery β€” TF-IDF vectorization + KMeans clustering (up to 8 themes) with guaranteed sentiment diversity
  • AI Summaries β€” OpenAI tool-use API generates labeled themes and an executive brief
  • Rating Trends β€” Monthly average rating and review volume charts via Recharts
  • Guardrailed Chat β€” 3-layer guardrail (regex pre-filter, system prompt boundary, hallucination scan) keeps responses grounded in review data
  • RAG Retrieval β€” pgvector cosine similarity search over 384-dimensional sentence embeddings
  • Real-time Progress β€” Server-Sent Events stream pipeline status to the frontend
  • Review Browser β€” Sortable, paginated review table with sentiment badges and star ratings

πŸ—οΈ Tech Stack

Layer Technology
Backend Python 3.11, FastAPI (fully async), Alembic migrations
Database Supabase PostgreSQL + pgvector
Embeddings all-MiniLM-L6-v2 via sentence-transformers (local, free)
LLM GPT-4o-mini via OpenAI API
Sentiment VADER (local, no API cost) with star-rating override
Clustering scikit-learn TF-IDF + KMeans
Frontend Next.js 14, Tailwind CSS, Recharts
Deployment Render (backend Docker) + Vercel (frontend)

πŸ”§ Getting Started

# Backend
cd backend && python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env  # fill in DATABASE_URL and OPENAI_API_KEY
alembic upgrade head && uvicorn app.main:app --reload

# Frontend
cd frontend && npm install
cp .env.local.example .env.local  # set NEXT_PUBLIC_API_URL
npm run dev

Open http://localhost:3000 to start analyzing reviews.


πŸ“¦ What's Changed

  • Update README with Vercel v0 transition note by @MomenMushtaha in #1
  • Update README with Vercel v0 transition details by @MomenMushtaha in #2
  • feat: async OpenAI, tests, metadata β€” Claude Code is back by @MomenMushtaha in #8
  • Improve reviews UX, theme analysis, and home page by @MomenMushtaha in #9
  • chore(deps): bump python-multipart from 0.0.12 to 0.0.22 in /backend by @dependabot[bot] in #10
  • chore(deps): bump next from 14.2.35 to 15.5.10 in /frontend by @dependabot[bot] in #11

πŸ†• New Contributors

Full Changelog: https://github.com/MomenMushtaha/ReviewLensAI/commits/v1.0.0