This is a submission for the Open Source AI Challenge with pgai and Ollama

What I Built

I built an intelligent bookmark management system for organizing URLs. The app uses a Chrome extension to save bookmarks into a web application.

Features include:

• Tags are automatically generated for bookmarks using pgai similarity analysis
• Using embeddings generated by Ollama's all-minilm model, searching is more intuitive
• Bookmarks can be further saved into collections
• Users can share bookmarks secure tokens

Demo

Project link
Code link

Tools Used

pgvector

• To store and index embeddings generated from bookmark content
• To enable semantic search functionality by finding similar bookmarks based on content meaning

pgai

• Used similarity functions to generate relevant tags
• Enhanced searching through AI-powered content understanding

Technology Stack

• Frontend: React.js
• Backend: Node.js, Express
• Database: TimescaleDB PostgreSQL
• AI Services: Ollama for embedding generation
• Authentication: Auth0

Final Thoughts

The most challenging but rewarding aspect was working on the automatic tagging system using pgai's similarity functions. Ideally, my demo should include a link to my Chrome extension, but due to time constraints, I am still in the process of submitting the extension for review.

Prize Categories

This submission qualifies for:

• Vectorizer Vibe
• Open-source Models from Ollama