DEV Community

Alex Kernel
Alex Kernel

Posted on

DiscovAI Search — Open‑Source AI Search Engine for Tools, Docs, and Custom Data

#webdev #programming #ai #openai

DiscovAI Search — Open‑Source AI Search Engine for Tools and Knowledge Bases

DiscovAI Search is an open‑source, AI‑powered search engine designed to index, understand, and search AI tools and custom knowledge bases using modern vector search + LLM reasoning.

The project combines:

  • semantic search (embeddings)
  • fast caching (Redis)
  • structured storage (Supabase / PostgreSQL)
  • modern frontend (Next.js)

It is suitable both as:

  • a production-ready AI search layer
  • an educational reference project for AI + web developers

GitHub repository:

https://github.com/DiscovAI/DiscovAI-search/

What Problem Does DiscovAI Search Solve?

Traditional keyword search fails when:

  • queries are vague or natural language
  • data is unstructured
  • users don’t know exact keywords

DiscovAI Search solves this by using semantic vector search, allowing users to search by meaning, not exact words.

Examples:

  • “tools for image generation”
  • “AI for document summarization”
  • “open-source alternatives to ChatGPT plugins”

Key Features

  • 🔍 Semantic Search with Embeddings
  • 🧠 LLM‑powered answer generation
  • ⚡ Redis caching for fast responses
  • 🗄️ Supabase (PostgreSQL + pgvector) backend
  • 🌐 Next.js frontend
  • 🔓 Fully open source

SEO keywords naturally covered:

AI search engine, vector search, semantic search, OpenAI embeddings, LLM search, Next.js AI app

High‑Level Architecture 

User Query
   ↓
Next.js API Route
   ↓
Embedding (OpenAI)
   ↓
Vector Search (Supabase / pgvector)
   ↓
Redis Cache (optional)
   ↓
LLM‑generated response
   ↓
UI
Enter fullscreen mode Exit fullscreen mode

This design keeps the system:

  • scalable
  • modular
  • easy to extend with new data sources

Tech Stack

  • Frontend: Next.js (React)
  • AI Models: OpenAI (embeddings + completion)
  • Database: Supabase (PostgreSQL + pgvector)
  • Cache: Redis
  • Language: TypeScript

Installation Guide (Local Setup)

Prerequisites

  • Node.js 18+
  • npm or yarn
  • OpenAI API key
  • Supabase account
  • Redis instance (local or cloud)

1. Clone the Repository 

git clone https://github.com/DiscovAI/DiscovAI-search.git
cd DiscovAI-search
Enter fullscreen mode Exit fullscreen mode

2. Install Dependencies 

npm install
# or
yarn install
Enter fullscreen mode Exit fullscreen mode

3. Environment Variables

Create a .env.local file:

OPENAI_API_KEY=your_openai_key

NEXT_PUBLIC_SUPABASE_URL=your_supabase_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key

SUPABASE_SERVICE_ROLE_KEY=your_service_role_key

REDIS_URL=redis://localhost:6379
Enter fullscreen mode Exit fullscreen mode

4. Supabase Setup

In Supabase:

  1. Enable the pgvector extension
  2. Create tables for:
    • documents
    • embeddings
  3. Store vectors as vector columns

This enables fast semantic search.

5. Run the App 

npm run dev
Enter fullscreen mode Exit fullscreen mode

Open:

http://localhost:3000
Enter fullscreen mode Exit fullscreen mode

You should now see the DiscovAI Search interface.

Indexing Data

DiscovAI Search can index:

  • AI tools
  • documentation
  • articles
  • internal knowledge bases

Typical flow:

  1. Add documents to Supabase
  2. Generate embeddings
  3. Store vectors
  4. Query via UI

Customization Ideas

  • Add your own datasets
  • Swap OpenAI for open‑source embedding models
  • Connect multiple vector indexes
  • Add authentication
  • Deploy to Vercel

Deployment

Recommended:

  • Frontend: Vercel
  • Database: Supabase
  • Cache: Upstash Redis

The project is cloud‑native and deploys cleanly.

Why This Project Is Useful

DiscovAI Search is valuable because it:

  • shows a real‑world AI search architecture
  • combines LLMs with vector databases correctly
  • is easy to fork and customize
  • works as both product and reference implementation

It’s a solid starting point for anyone building:

  • AI‑powered search
  • internal knowledge assistants
  • tool discovery platforms

Final Notes

This project demonstrates how modern AI search systems are built today, not in theory.

If you’re interested in:

  • semantic search
  • vector databases
  • LLM‑powered UX

DiscovAI Search is worth exploring.

Top comments (0)