llmware

Why we included this project

llmware suits teams that want to ground an LLM in their own documents without assembling a pipeline from separate tools. It handles the whole retrieval path in one library: parsing PDFs, Office files, and HTML, chunking the text, embedding it into a vector store, and searching with text, semantic, or hybrid queries. The bundled model catalog is what stands out, with 300+ quantized models and a set of small, task-specialized ones tuned for enterprise workflows, all reachable through the same interface and runnable on a laptop or edge device. That makes it a solid choice for private, on-premise answers, though you can still call OpenAI, Anthropic, or Google models when a cloud API makes more sense. The library also keeps multiple knowledge bases organized per project, which helps when you manage several document collections at once.