| title | Astoria Open G |
|---|---|
| emoji | ⛵ |
| colorFrom | blue |
| colorTo | green |
| sdk | docker |
| pinned | false |
An advanced, multi-LLM application that allows users to query a maritime history database using natural language.
- Natural Language to SQL: Ask complex questions in plain English and get answers directly from a PostgreSQL database.
- Intelligent Routing: Automatically uses a fast, direct chain for simple queries and a powerful Gemini-powered agent for complex, analytical questions.
- RAG Pipeline: Enriches SQL data with contextual information from a vector database to provide comprehensive, narrative summaries.
- Multi-LLM Architecture: Leverages Google's Gemini for high-accuracy SQL generation and Groq for high-speed narrative synthesis.
- Check the External Service Status panel to ensure all systems are operational.
- Type a question about the maritime database into the text box (e.g., "List 10 vessels built after 1900").
- Click Submit to see the generated SQL, the raw data results, and a synthesized narrative summary.
- Backend: FastAPI, Uvicorn, LangChain
- Frontend: React, Vite
- LLMs: Google Gemini, Groq
- Database: PostgreSQL with pgvector (via Supabase)