Qdrant can now embed raw text during upsert through Cloud Inference, and qdrant-client can embed client-side with FastEmbed. It still does not crawl a Git repository, parse Markdown, choose chunks, or reconcile changed and deleted files.

What Qdrant itself now removes from the ingestion pipeline

Qdrant can perform inference at write and query time. Qdrant Cloud Inference lets an application send text through the Qdrant API and have Qdrant generate and store vectors; the Python client also integrates FastEmbed for client-side embedding.

This means a custom indexer does not necessarily need a separate embedding service or embedding SDK. It can send chunks as text plus metadata and let the Qdrant integration produce vectors.

The boundary remains important: current Qdrant documentation does not expose a repository or directory ingestion primitive that recursively reads files, parses Markdown, chooses chunk boundaries, tracks Git state, or removes stale chunks after deletion or rename. Qdrant starts after the application has selected the content to index.

Qdrant's own data-management integrations list external ingestion tools such as Unstructured, Chonkie, Airbyte and others. That separation is consistent with Qdrant being the vector/search layer rather than a Git-aware document loader.