DiscovAI-Search provides an AI-powered search engine capable of indexing both a curated database of AI tools and user-provided custom data. It targets developers and researchers seeking efficient ways to discover AI tools or query private datasets using natural language. The primary benefit is leveraging advanced AI techniques like vector search and LLMs to deliver precise, context-aware search results and answers, significantly enhancing information retrieval efficiency.

How It Works

The engine employs a hybrid approach combining vector-based similarity search with LLM-driven response generation. User queries are converted into vector embeddings using Jina AI models, enabling precise similarity matching against indexed AI products or custom data stored in Supabase with Pgvector. Search results are cached using Upstash Redis to accelerate responses for repeated queries. Finally, GPT-4o-mini synthesizes information from retrieved results to generate detailed, context-aware answers, enhanced by text streaming via the Vercel AI SDK.

Quick Start & Requirements

• Install: Clone the repository and install dependencies using pnpm i.
• Prerequisites:
  • Node.js environment with pnpm.
  • Supabase project setup with src/db/init.sql executed.
  • Upstash Redis database instance.
  • API Keys: Jina AI (JINA_API_KEY), OpenAI (OPENAI_API_KEY, OPENAI_API_URL), Supabase (NEXT_PUBLIC_SUPABASE_URL, NEXT_PUBLIC_SUPABASE_ANON_KEY), Upstash Redis (UPSTASH_REDIS_REST_URL, UPSTASH_REDIS_REST_TOKEN).
  • Configuration via .env.local file.
• Run: Execute pnpm dev to start the local development server.
• Links: Live Demo available at DiscovAI.io.

Highlighted Details

• Utilizes gpt-4o-mini for LLM responses and Jina AI for embeddings.
• Employs Pgvector within Supabase for efficient vector storage and search.
• Features Redis caching via Upstash for performance optimization.
• Built with Next.js, Tailwind CSS, shadcn/ui, and Radix UI primitives.

Maintenance & Community

The project encourages community interaction via GitHub issues. Specific details regarding core maintainers, active development, or community channels (like Discord/Slack) are not detailed in the provided README.

Licensing & Compatibility

The license type is not explicitly stated in the provided README. Users should verify licensing terms before commercial use or integration into closed-source projects.

Limitations & Caveats

Setup requires integrating multiple external services (Supabase, Upstash) and obtaining several API keys, which may present a barrier to entry. The project's reliance on these external services means their availability and pricing models directly impact DiscovAI-Search's functionality and cost. The current README does not specify performance benchmarks or known limitations beyond setup requirements.