E-commerce will evolve through four levels with AI: from filling out forms to anticipating needs. But when AI agents do all the research, it could create perfect competition and squeeze out products with soul.
Podcast: Ben Thompson from Stratechery on AI ads, the end of SaaS, and the future of media
What happens when AI starts being integrated into your online shopping experience?
Stripe cofounder John Collison describes four levels:
- Just-in-time UI: AI fills out web forms for you.
- Parameter-based search: AI replaces keyword searches with specific queries, such as "a jacket rated for minus ten degrees in Tokyo next week."
- Anticipatory commerce: AI profiles your preferences and recommends things before you search.
- Perfect matching: AI finds a hyper-specialized product from a random seller anywhere in the world and matches it to your exact need.
Here's an implied structural risk: If AI agents do all the research, every product gets compared purely on measurable specs.
The unmeasurable qualities, the craft and soul, fall by the wayside. More information can mean better decisions, but it can also flatten everything.
I also like Zuckerberg's perspective that "Meta's ad system is the largest and most successful agent in the world today" because it can match specialized products with people who didn't even know they wanted them.
It's also interesting to use the lens to understand why Meta chose to acquire Manus.