Why fact extraction and published claims stay separate
Abstracts
Abstracts
One design choice behind ProfileScribe is deliberate: the crawler agent that fetches approved public links only extracts timestamped facts. It does not write the sentences that end up on a profile. That separation exists because professional writing has a habit of quietly overstating what evidence actually supports, especially when the writer is drafting about their own work.
So raw extraction feeds a source graph, and any claim built from it still passes through a review queue before it is publishable. It is a slower path, but it is the difference between a profile that reflects what happened and one that reflects what sounded good. For durable professional memory, that distinction matters more than speed.