Catholic Knowledge Base Project

write a short post about this and my project I made

Over the past few weeks I've been working on something I call the **Catholic Knowledge Base**—a local-first AI system designed to answer questions using official Catholic sources rather than generic internet content. Instead of relying on whatever an LLM happens to remember, the system retrieves information directly from the Catechism, papal encyclicals, Church Fathers, councils, canon law, and other approved Catholic documents, then grounds its answers in those sources. The goal is simple: create an AI assistant that behaves more like a research librarian than an oracle. When someone asks a question such as *"What is good?"*, the system searches a curated corpus of Catholic teaching, retrieves the most relevant passages, and synthesizes an answer while showing exactly where the information came from. This makes the reasoning process transparent and allows users to trace claims back to authoritative documents. What makes the project interesting to me is that it combines several ideas I've been exploring for a long time: local-first AI, retrieval-augmented generation (RAG), knowledge organization, and sovereignty over information. Rather than sending data to a black-box service, the entire knowledge bank can be owned, inspected, expanded, and customized by the community. The screenshot above shows the system operating over more than 200 Catholic documents and thousands of indexed knowledge chunks, allowing it to move seamlessly between scripture, magisterial teaching, catechetical references, and historical sources. In many ways, this project is an experiment in building AI systems that are accountable to a tradition of knowledge. Instead of asking an AI what Catholicism teaches, the AI is trained to ask the Church's own documents first. The result is a tool that can help with study, catechesis, research, and exploration while keeping the original sources at the center of the conversation.