Register a callback to receive progress updates during pipeline execution.
Process a content source (websites, files, etc.) through the pipeline. Loads content items from the source, then vectorizes each one.
Process an entire entity through the knowledge pipeline.
Stages executed depend on the params:
Process a single record through the pipeline.
Orchestrates the full knowledge pipeline: ingest -> extract -> autotag -> vectorize.
This class coordinates between the autotagging engine and the vector sync engine to produce both human-readable tags and machine-readable embeddings in a single pass.