Summary

This project offers a personal knowledge base AI Q&A system built upon Retrieval-Augmented Generation (RAG) principles and the Spring AI framework. It targets developers and power users aiming to construct custom, context-aware AI assistants capable of querying specific knowledge repositories and handling multi-modal inputs, serving as a practical learning resource for Spring AI and RAG integration.

How It Works

The system leverages Spring AI 1.x to manage Large Language Model (LLM) interactions and RAG workflows. It integrates PGVector as a vector database for efficient semantic search over document embeddings, enabling granular knowledge base separation through custom query conditions. Conversation history is persisted using a custom DatabaseChatMemory implementation. Multi-modal support includes storing attachments like documents and images in Minio, with planned enhancements for extracting and describing embedded images to enrich conversational context.

Quick Start & Requirements

• Primary Install/Run: Frontend: pnpm install followed by pnpm start. Backend: Modify application.yml and llm.yml (adjusting spring.config.import as needed), then start SystemApp.
• Prerequisites: Node.js v18, JDK 17. Requires Minio and PGVector, typically launched via the provided docker-compose.yml. The system is designed to interface with OpenAI-compatible APIs, potentially via a proxy like One-API.
• Setup: Requires Docker for essential services (Minio, PGVector). Backend configuration adjustments are necessary.

Highlighted Details

• Implements RAG using the latest Spring AI 1.0 version.
• Features knowledge base separation, allowing targeted RAG queries.
• Supports multi-modal input with attachment storage in Minio.
• Conversation history is managed via a custom DatabaseChatMemory.

Maintenance & Community

No specific details regarding project maintainers, community channels (e.g., Discord, Slack), roadmap, sponsorships, or partnerships were provided in the README snippet.

Licensing & Compatibility

The license type and any associated restrictions for commercial use or integration with closed-source projects are not specified in the provided text.

Limitations & Caveats

The project is explicitly positioned as a learning example. Core functionalities such as full multi-modal RAG and automated image extraction/description from documents remain as TODO items, indicating they are not yet implemented.