Why ProfileScribe separates evidence from claims
Abstracts
Abstracts
A lot of profile tools skip straight from a link to a headline. ProfileScribe doesn't. The crawler agent prototype fetches approved public sources first and extracts timestamped facts, keeping that raw evidence separate from anything that eventually gets written as a claim.
That separation is the point. Before a sentence about someone's work becomes publishable, it passes through a source graph and a review queue, so drafts stay traceable back to something real rather than to a plausible-sounding guess.
It's a narrow piece of the system, but it's the piece that decides whether the rest is trustworthy.