What is agentic commerce? The definitive guide for retailers

Agentic commerce is the most significant shift in how products get discovered and bought since mobile shopping. For two decades, retailers optimised for human attention — hero images, brand campaigns, retargeting, and checkout flows designed for thumbs. That model still matters. But a parallel channel is opening: AI agents that read your data, verify whether they can trust it, and decide whether to recommend — or buy from — you.

The short definition

Agentic commerce is when AI systems — not human shoppers — handle discovery, comparison, trust verification and purchase completion on behalf of a customer. The agent does not care about your homepage. It reads your product feed, your schema markup, your API responses, and your checkout surface. If the data is missing, stale or inconsistent, you are invisible or disqualified.

Why this is happening now

Three forces converged in 2024–2026. First, large language models became reliable enough to act as shopping intermediaries — understanding intent, comparing options, and explaining trade-offs in natural language. Second, payment networks and platforms began standardising agentic checkout protocols (Google's Universal Commerce Protocol, Visa's Trusted Agent Protocol, PayPal agentic commerce tools). Third, consumers started using AI for high-intent purchase research, especially in categories where specification, availability and delivery date matter as much as brand.

Global signals retailers should track

• OpenAI and Shopify partnered on agentic shopping flows for merchant catalogues.
• Google launched Universal Commerce Protocol (UCP) for agent-mediated checkout.
• Amazon's Rufus and Buy for Me features push agent-assisted purchasing inside its ecosystem.
• PayPal, Visa and Mastercard are building trust and payment rails for autonomous agents.
• McKinsey estimates agentic AI could automate 60–70% of current work activities; commerce is a primary use case.
• Adobe's 2026 commerce predictions highlight agentic discovery as a top retailer priority.

What agents actually read

An AI buying agent does not experience your brand the way a customer does. It evaluates machine-readable signals. The minimum viable surface includes product identifiers (GTIN/EAN), accurate price and currency, live availability, delivery cut-offs and promises, structured reviews, return policy, and a checkout path it can complete. Missing any critical field does not mean a poor experience — it means elimination from the consideration set.

The two-layer readiness model

Most retailers only think about the storefront. Agentic commerce requires equal attention to the outside layer (what AI buyers see) and the inside layer (how your operation runs). A perfect product feed means nothing if orders, inventory and logistics are disconnected. A unified commerce stack means nothing if agents cannot read your catalogue in the first place.

1. Found — structured feeds, schema, marketplace surfaces agents can parse.
2. Trusted — accurate, fresh, consistent data across every channel an agent checks.
3. Bought — checkout and payment surfaces agents can complete.
4. Run — unified commerce: orders, warehouses, logistics, PIM connected.
5. Governed — clear rules for what to automate, review, or ignore.

What UK retailers should do this quarter

Start with an honest audit. Score your estate across agent discoverability, product data integrity, feed freshness, checkout readiness, API surface, and operational automation. Prioritise fixes by commercial impact — not by which vendor launched a press release this week. Most mid-market UK retailers can close the first critical gaps with existing teams once they know exactly what agents see.

For the first time in twenty years, the field is level. An AI agent cannot be impressed by your ad budget. It reads your data, checks whether it can transact, and decides.

— Beyond Partners