Keeping fact extraction separate from the claims a profile makes
Abstracts
Abstracts
One design choice in the crawler agent prototype is a strict separation: the crawler's job is to fetch approved public links and extract timestamped facts, nothing more. Turning those facts into a claim a person would publish is a separate step, handled with more scrutiny.
It's a small architectural decision, but it matters for trust. If evidence gathering and claim-making happen in the same step, it's easy for a system to overstate what a source actually shows. Keeping them apart makes it possible to check, at any point, whether a stated fact traces back to something real.