This is a submission for the Redis AI Challenge: Real-Time AI Innovators.
What I Built
BetterReviews is an intelligent platform that empowers users to make informed purchasing decisions through AI-driven analysis. The platform enables users to share authentic product reviews and experiences while providing comprehensive product insights to potential buyers.
Key Features
Comprehensive Product Analysis - For any product inquiry, BetterReviews delivers detailed analysis including key specifications, standout features, known limitations, target audience, and competitive comparisons.
AI-Driven Review Insights - Leverages user-generated reviews from the platform to extract meaningful insights, helping buyers understand real user experiences, common issues encountered, and the aspects users appreciated most.
Tech Stack
Flask - Web application framework for API development and frontend serving
Redis - Primary vector database for efficient storage and retrieval of product reviews and embeddings
RedisVL - Redis vector library managing index operations, vector search capabilities, embedding generation, and semantic caching
LangChain - Integration framework connecting Google Gemini LLM with the RAG pipeline for intelligent review analysis
Challenges & Learnings
This was my first AI agent project, so concepts like vector databases and RAG were completely new. Redis's detailed documentation and tutorials made the learning curve really smooth and efficient.
Getting the LLM to generate consistent, desired and quality analyses took lots of trial and error. I learned that clear context and structured output formats make all the difference.
Overall, building this deepened my understanding of modern AI applications and was genuinely fun to work on!
How I Used Redis 8
Vector Database - Redis serves as the primary vector database, storing product review embeddings for efficient similarity search and retrieval.
Data Storage - Each product review is stored as a Redis hash, making it easy to manage and query structured review data alongside vector embeddings.
Vector Search - Leveraged RedisVL library to perform vector searches, retrieving the most relevant product reviews to provide as context for the LLM's insight generation.
Semantic Caching - Implemented RedisVL's semantic cache to store results for previously queried products, significantly reducing response times and avoiding redundant processing for repeat queries.
Demo
Demo Video Link - BetterReviews
GitHub Repo - GitHub Repo
Top comments (0)