Knowledge

What happens after upload

How sources are processed and stored for retrieval.

After you upload a source, Spaces prepares it for retrieval. The details panel shows exactly what happened.

Processing stages

  1. Ingestion — the file is parsed and text is extracted.
  2. Chunking — content is split into segments sized for search (size and overlap are set in System).
  3. Embedding — each chunk is converted into a numerical representation used for search.
  4. Indexing — chunks, embeddings, and metadata are stored in your Space for retrieval.

What the details panel shows

Open a source's details to see:

  • Processing status and dates
  • Original file size (original is not stored)
  • Number of chunks created
  • Storage breakdown: content, metadata, embeddings, approximate total

What is retained

Original files are not retained. Only the indexed segments, search representations, and metadata needed for retrieval are stored inside your Space perimeter. Your sources stay private to your Space and exist to ground answers from the Knowledge you added.

For the full pipeline in plain language, see How Spaces works.

Organize and maintain

Keep your Knowledge accurate over time by finding, refining, reusing, and removing sources.

Ask questions

Start a chat, read streaming answers, and see how each response was produced.