Show HN: IA-SQL – Postgres compiles your documents into a wiki with an LLM (github.com)

IA-SQL has emerged as an innovative tool that transforms PostgreSQL into a self-compiling knowledge base, implementing Andrej Karpathy's "LLM Wiki" concept. By allowing users to insert raw documents directly into the database, IA-SQL employs a background worker that utilizes an external language model (LLM) to compile these documents into a structured, cross-referenced Markdown wiki. This design shifts the complexity of knowledge integration from query time to ingestion time, ensuring that the wiki evolves and improves with each new document, thereby enhancing overall knowledge consistency while auditing for inaccuracies or "hallucinations." For the AI/ML community, this development is significant as it represents a departure from traditional chat-based retrieval systems that typically operate using Retrieval-Augmented Generation (RAG). Instead of reconstructing context during every query, IA-SQL intelligently compiles knowledge, allowing it to grow organically and remain coherent. The integration of PostgreSQL's architecture for managing background processes and triggers facilitates a stable, asynchronous environment that supports LLM operations without risking the integrity of the database. As an evolving proof of concept built on PostgreSQL 17, IA-SQL underscores the potential for databases to actively contribute to knowledge management and enhances accessibility to combined insights across diverse data sources.