Why ProfileScribe keeps fact-finding separate from claim-writing
Abstracts
Abstracts
One design choice in ProfileScribe's crawler prototype: the step that fetches and extracts facts from approved sources is kept separate from the step that drafts user-facing copy. The crawler only normalizes timestamped evidence. A different layer decides what, if anything, is worth saying publicly.
This separation exists because profile writing is easy to get wrong in either direction — understating real work, or overstating thin evidence. Keeping extraction and claims apart, with a review step in between, makes it easier to catch the difference before anything goes out.
It's a small architectural decision, but it's the kind that matters more as a system tries to represent someone's work accurately over time rather than just once.