insights · what we are learning ·
What is agentic commerce? A plain guide for brands
Agentic commerce is shopping done with, or for, a person by AI software that can search, compare and in some cases pay. Instead of typing into a search box and opening ten tabs, the shopper tells an assistant what they want, and the assistant decides which brands to name. For a brand, that means being readable by machines now matters as much as being appealing to people.
What is agentic commerce?
Shopping where an AI agent does part of the work. An agent is software that doesn't only answer a question but takes steps: it searches, compares, fills a basket and, if the shopper allows it, pays.
It sits on a ladder, and most brands are on the bottom rung:
- Answers.The shopper asks ChatGPT, Gemini, Claude or Meta's Muse what to buy and gets a short list of brands. Nothing is bought inside the assistant.
- Apps.A brand has its own app inside the assistant, so the shopper can browse its range without leaving the conversation. Prada launched a ChatGPT pilot on 28 October 2026, and Louis Vuitton showed one in June.
- Agents that buy.The assistant picks a product and completes the checkout, within limits the shopper sets. Google's Gemini and AI Mode do this in the US through UCP. Meta's Muse does it with a one time card from Stripe's Link.
How does it work?
An agent can only recommend or buy what it can read. It reads three things: your product data, your site's markup, and the standards that let it talk to a shop directly.
Two of those standards are worth knowing by name. UCP, the Universal Commerce Protocol, is Google's, announced in January 2026. It lets an agent see a merchant's catalogue, prices and stock, and start a checkout. MCP, the Model Context Protocol, is an open standard that lets an assistant use a shop's tools, such as searching its range. A brand with neither is still read, but only by scraping its pages, which agents do badly.
In our research of 75 UK brands, every site that had both UCP and MCP ran on Shopify. Everyone else was invisible to the agents that need them.
Why does it matter?
Because it's where shopping starts. In a survey of US consumers by L.E.K. Consulting, reported by MarTech in July 2026, 46% of people who use AI started their purchase research on a standalone AI platform, up from 25% in 2024. Traditional search fell from 43% to 24% as a starting point.
The Gemini app passed a billion monthly users in August 2026.
The visitors who do arrive are worth more. Adobe measured AI referral traffic to US retail sites up 62% year on year in July 2026, converting 60% better than other traffic.
Is anyone buying this way?
Yes, and it's growing fast, most of all in the US, where agents have been live longest. In Adobe's survey of over 5,000 US consumers, reported in April 2026, 39% already use AI for online shopping and 85% of them said it improved their experience. AI traffic to US retailers rose 393% year on year in the first quarter of 2026.
The UK is earlier and moving the same way. Adyen's 2026 retail report found the share of UK shoppers using AI for shopping rose from 12% to nearly 28% in a year, and 44% said they'd let AI handle a whole purchase once they'd set a budget and criteria.
Checkout is still taking shape. Google and Meta check out inside their assistants in the US, and since March 2026 ChatGPT sends shoppers to the retailer's own checkout, with brands building apps there. Our read: discovery moved first, and the buying is following it.
Who is building it?
Everyone with an assistant. OpenAI, Google, Meta, Microsoft and Perplexity all let shoppers buy or are building toward it. Apple said in January 2026 that its next foundation models would be based on Google's Gemini. Amazon is building its own assistant and keeping the others out.
That's why we don't treat this as a bet on one platform. A brand that is readable by agents is readable by all of them. And if a few assistants do come to dominate, regulators have a long record of insisting on fair access, which we think they will here too: our read on how.
What should a brand do about it?
Four things, in order.
- Find out what the assistants say about you.Ask the questions your shoppers ask, in each market you sell in. The free score does this across eight engines.
- Treat the product feed as brand work.Titles, variants, prices and stock, kept current, with descriptions that answer questions. It's the first thing an agent reads.
- Give agents an address.A UCP profile and an MCP address let any agent search your range and get live price and stock. If you're not on Shopify, Storefront does this from your product feed.
- Own the relationship after the sale.Accounts, email and a community the customer chooses to join. An assistant can win the first visit. The brand has to win the second.
Where does this leave brands?
In the same place as ever, with one new first impression to manage. People still buy from brands they trust. The difference is that the introduction now often comes from a machine reading your data. Your next customer is AI. Your best is still human.
Questions
What is agentic commerce in one sentence?
Shopping in which AI software searches, compares and sometimes pays on the shopper's behalf.
How is it different from AI search?
AI search answers a question. An agent acts on the answer: it builds a basket, starts a checkout or keeps watching a price.
Are shoppers using it?
Yes, and quickly. In Adobe's survey of over 5,000 US consumers, 39% already use AI for online shopping and 85% of them said it improved their experience. In the UK, use rose from 12% to nearly 28% in a year, and 44% would let AI handle a whole purchase once they'd set the criteria.
Does a brand need to be on Shopify?
No. In our research of 75 UK brands every site with both UCP and MCP ran on Shopify, which is the gap Storefront fills for brands that aren't.
How do I know if my brand shows up in AI answers?
Ask the assistants the questions your shoppers ask, in each market you sell in, or run the free score, which does it for you.