Aggregation and AI
The aggregator playbook was built on a simple insight: own the user relationship, commoditize supply, and let demand do the rest. Google owned the query, Facebook owned the feed, and everyone else lined up to be indexed or ranked.
Agents break the premise. When the interface is a model acting on your behalf, the "user relationship" belongs to whoever the agent trusts — and the agent, unlike the user, reads the fine print.
The scarce resource is no longer attention. It is authorization.
Consider email, the oldest aggregation surface of all. An inbox aggregates everyone who wants something from you; the sorting was always yours to do. An agent that sorts it — that decides what deserves your attention and what never will — inverts thirty years of assumptions about who owns distribution.
Which raises the question every SaaS founder should be asking this year: when your customer's agent answers the door, what does your onboarding email even mean?
Matt Levine on whether an index can commit securities fraud, plus SPACs, again…
2h ago · 15 min