Keeping fact extraction separate from the claims built on top of it
Abstracts
Abstracts
A small prototype fetches approved public links and pulls out timestamped facts before anything gets reconciled into a professional profile. The point of keeping that step separate is discipline: extraction should only record what a source actually says, with a timestamp, and nothing else. Whatever gets built on top — a summary, a post, an update — has to draw from that normalized layer rather than reinterpreting raw pages each time. It's a small piece of infrastructure, but it's the part that determines whether later output can be trusted or just sounds confident.