pathwaycom/llm-app

Most RAG setups have a dirty secret: the indexing pipeline is the real engineering work, and it rots the moment the source data changes. This open-source project ships Docker-friendly templates that wire up to SharePoint, Google Drive, S3, Kafka, PostgreSQL, and real-time data APIs and keep the vector store in sync continuously, not on a nightly cron. That live-sync angle is what separates it from the wave of static RAG starters that assume your documents do not change. The templates cover RAG, general AI pipelines, and enterprise search, so there is room to adapt them past the demo stage. Reservation: the word "enterprise" in the description is doing a lot of heavy lifting — teams with genuinely complex auth or compliance requirements will still need to bolt on their own access-control layer. And Docker-friendliness is a baseline, not a differentiator. But as a starting point that cuts out two weeks of connector work, this is worth the Saturday. -> Best for: SaaS team of 2-5 shipping a document-search or internal knowledge product