Shopping agents are coming. Is your retail architecture ready to be read by a machine?
Retail leaders expect AI shopping agents to reshape their business, but the data tells a more nuanced story, and an architectural one.
Expectation is running ahead of usage
- In BCG’s October 2026 research, 68% of retailers expect agentic commerce to have a significant or transformational effect within two years. BCG projects that more than half of e-commerce spending could be agent-assisted by 2030.
- Yet agent referrals made up less than 1% of website visits in late 2025, and only 4% of consumers would let an agent complete a purchase on their own.
- On 30 October, DoorDash disclosed it had received only “several hundred” orders from third-party AI agents, against millions of orders a day.
Readiness is the real gap
Only 18% of retailers in BCG’s survey consider their marketplaces ready for agents. 91% name losing the customer relationship as their top worry. Meanwhile, the payment networks are building the trust layer. Visa has announced Agent Scoring and an Agentic Registry to identify agents, and Mastercard’s Agent Pay work focuses on agent identity, verifiable intent and auditable records.
What this means for the enterprise architecture
- Make product data machine-readable first. Agents compare on structured facts. Clean, rich product attributes, real-time prices, live inventory and accurate fulfilment promises, published through feeds and APIs, are the foundation. In BCG’s survey, 67% of retailers plan to enrich product data and 58% plan to provide real-time availability within two years.
- Separate what you share from what you protect. Share product and availability data openly with agents. Keep customer identity, preferences and behavioural data behind your own walls. Design this as one access policy enforced at an API gateway, not as case-by-case deals.
- Recognise and manage agent traffic. Identify agents at the edge, apply rate limits and terms, and decide which agents are welcome. Plan to accept agent identity and intent signals from payment networks as they mature.
- Express loyalty in terms agents can compare. Member prices and delivery guarantees should be exposed as data, while exclusive experiences stay in your own channels to give shoppers a reason to come direct.
- Keep checkout and the customer relationship in your hands. Most agent experiences still hand off to the retailer’s checkout. Make that hand-off fast and reliable, and keep the order, the receipt and the after-sales journey on your platform.
The takeaway
Agentic commerce is early, so this is the right moment to prepare, not panic. Retailers that make their catalogue easy for machines to read, while keeping customer data and checkout firmly in their own hands, will win whichever agents shoppers end up using.
KSL helps retailers and consumer brands design API-first, agent-ready commerce architectures. Next week in this series, we look at another industry.