ReviewMind

ReviewMind is a production-style review intelligence dashboard for marketing teams. It turns public review data into evidence-backed insight across sentiment, recurring themes, competitor gaps, and messaging opportunities.

The product is designed to feel calm, minimal, and premium rather than like a generic AI dashboard.

What ships in this MVP

• Minimal landing and onboarding flow
• Multi-page dashboard shell with sidebar and top bar
• Overview, Sentiment, Themes, Competitors, Messaging, Evidence, and Settings views
• FastAPI review ingestion pipeline
• Google Play Store collection support via google-play-scraper
• SQLite persistence for analysis runs and processed review rows
• Sentiment analysis with VADER-style scoring and rating-aware calibration
• Human-readable theme extraction
• Evidence-backed marketing recommendations
• Clearly labeled demo fallback when live review collection is unavailable

Project structure

.
+- apps/
¦  +- api/
¦  ¦  +- app/
¦  ¦  ¦  +- api/routes/
¦  ¦  ¦  +- core/
¦  ¦  ¦  +- services/
¦  ¦  ¦  +- utils/
¦  ¦  +- data/
¦  ¦  +- storage/
¦  ¦  +- requirements.txt
¦  +- web/
¦     +- app/
¦     +- components/
¦     +- lib/
¦     +- public/
+- docs/
+- .env.example
+- README.md

Setup

1. Backend

cd apps/api
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000

2. Frontend

cd apps/web
npm install
npm run dev

Open http://localhost:3000.

Environment variables

Copy C:\Users\hello\Desktop\Webapp\.env.example to .env and adjust values as needed.

Important values:

• NEXT_PUBLIC_API_BASE_URL: backend API base URL
• REVIEWMIND_DEFAULT_COUNTRY: Google Play country for search and review collection
• REVIEWMIND_DEFAULT_LANGUAGE: default review language filter
• OPENAI_API_KEY: optional, enables LLM-assisted summaries and copy suggestions
• OPENAI_MODEL: optional model override

How the analysis works

1. Collect public reviews from Google Play Store
2. Normalize fields and deduplicate review rows
3. Clean and language-filter text
4. Score sentiment
5. Assign usable marketing themes
6. Compare products across praise, complaints, and topic prevalence
7. Generate evidence-backed summaries and messaging recommendations
8. Store the run for later exploration and export

Notes on trust

• ReviewMind never invents live evidence.
• Quotes are only taken from extracted review rows.
• Confidence labels are tied to sample size, product coverage, and whether the analysis is demo data.
• If a run uses fallback data, the UI says so explicitly.

Future improvements

• App Store and Trustpilot ingestion
• CSV upload and scheduled refreshes
• richer topic modeling with BERTopic or embeddings
• stronger language detection and translation support
• saved workspaces and team collaboration
• exportable insight decks and share links
• Postgres + background jobs for larger workloads