Why fact extraction and published claims stay separate
Abstracts
Abstracts
One design choice in ProfileScribe: the crawler agent that reads approved sources never writes directly to a public profile. It normalizes what it finds into timestamped facts first, and only then does a separate step turn that evidence into a draft update, which sits in a review queue before anything is published.
The split matters more than it sounds. Professional profiles get unreliable when summarization and evidence collection happen in the same step, because it becomes easy to state something slightly ahead of what the source actually supports. Keeping extraction and claims apart, with a human review step in between, is a small structural decision aimed at keeping updates grounded rather than persuasive.