Re:What breaks when AI agents do the shopping? (200-page book)
by toddpiechowski on Wednesday February 11, 2026 @08:43AM
(#46974872)
One page summary of the book:
The “agentic commerce” stack:
1) Intent (what the user actually needs)
2) Research (what options exist)
3) Evaluation (tradeoffs + trust signals)
4) Transaction (permissions/auth/payment)
5) Fulfillment (delivery, returns, satisfaction loop)
What changes vs today:
- SEO/ads matter less if agents don’t browse pages
- structured product data + availability + policies matter more
- reputation signals become machine-readable and portable
- checkout becomes delegated + constrained (limits, identity, auditability)
If you want to argue against it, the best angles are:
- why humans will keep browsing (and for which categories)
- where trust signals will come from (and who controls them)
- whether retailers will allow agents at scale
The “agentic commerce” stack: 1) Intent (what the user actually needs) 2) Research (what options exist) 3) Evaluation (tradeoffs + trust signals) 4) Transaction (permissions/auth/payment) 5) Fulfillment (delivery, returns, satisfaction loop)
What changes vs today: - SEO/ads matter less if agents don’t browse pages - structured product data + availability + policies matter more - reputation signals become machine-readable and portable - checkout becomes delegated + constrained (limits, identity, auditability)
If you want to argue against it, the best angles are: - why humans will keep browsing (and for which categories) - where trust signals will come from (and who controls them) - whether retailers will allow agents at scale