When you upload files to the Library, they go through an indexing pipeline: content is extracted, split into chunks, converted into vector embeddings, and stored in a vector database (Qdrant). During conversations, agents can automatically search your indexed files for relevant chunks and include them in their responses — even without you explicitly referencing the files. This is especially powerful for technical documentation, specs, and reference material. Processing status is shown on each file — wait for indexing to complete before expecting the agent to find your content.