Short answer: Since we last wrote about agentic commerce, the landscape has shifted in four ways worth knowing. First, two competing protocols became three "camps" - Google's UCP, OpenAI and Stripe's ACP, and a dedicated agent-payments layer (AP2) - with shared plumbing like MCP underneath. Second, there's been a reality check: reports that OpenAI pulled back its flagship in-chat checkout after weak sales, even as the underlying protocol continues - a reminder that being present isn't the same as being ready. Third, "universal translator" middleware has emerged that lets merchants support all the protocols at once, which validates the case for not betting on one. And fourth - most important for UK retailers - it's arriving here. Through all of it, the underlying truth hasn't moved: clean, structured, machine-readable data is the foundation every protocol depends on.
This is a snapshot update to our earlier piece on what agentic commerce means for retail. The fundamentals there still hold; this covers what's changed on top.
Change 1: two protocols became three camps
In our last piece the story was ACP versus UCP. It's now better understood as three camps that increasingly interlock:
UCP (Universal Commerce Protocol) - Google and Shopify's open standard, public since January 2026, letting AI surfaces like Search AI Mode and Gemini discover merchants and orchestrate the full journey. It's endorsed by a wide coalition including major retailers and the card networks.
ACP (Agentic Commerce Protocol) - OpenAI and Stripe's standard, focused on running a full checkout inside an assistant.
AP2 (Agent Payments Protocol) - Google's payments layer that sits alongside UCP, handling the signed authorisation a network needs before an agent pays. Notably, AP2 uses W3C Verifiable Credentials - signed "Intent, Cart, and Payment" mandates that create a tamper-proof chain of permission, so an agent can prove it was authorised to buy.
Underneath all three sit shared standards - the Model Context Protocol (MCP) and agent-to-agent messaging - which is why the camps interoperate rather than fully compete. The practical takeaway: this is consolidating around shared infrastructure, not resolving into a single winner.
Change 2: a reality check on in-chat checkout
The most instructive development is a stumble. Industry trackers report that OpenAI quietly wound down its flagship in-chat Instant Checkout surface in early 2026 after it ran for several months with very low sales - while the ACP protocol itself continues. (Signals here are mixed, and assistant shopping features remain active in various forms, so treat the specifics as a fast-moving picture rather than a settled fact.)
Whatever the exact status, the lesson is the strategically useful part: a flashy consumer checkout button is not the same as a working agentic commerce capability, and early front-ends can fail even when the direction of travel is right. It's a near-perfect echo of the earlier chatbot and RAG hype cycles - the shiny surface cooled, the underlying foundation kept advancing. For retailers, it's a reason to invest in the durable layer (clean, structured data and protocol-readiness) rather than chase whichever assistant feature is generating headlines this quarter.
Change 3: "support everything" middleware validates the no-regret play
In our last piece we argued you shouldn't bet on a single protocol. The market has since made that easier - and proved the point. Payment platforms have begun launching universal "agentic" layers that support UCP, ACP, AP2, and other emerging standards simultaneously, positioning themselves as a single translator so merchants don't have to choose. Stripe, for its part, has backed more than one protocol; Shopify has made UCP effectively self-serve, letting developers register an agent and call a public endpoint without an approval gate.
The strategic implication is unchanged but strengthened: the winning move is to be reachable through all the relevant protocols, and the way you achieve that is by getting the shared foundation right - clean product data and an adapter approach - rather than committing to one ecosystem. The middleware handles the translation; you handle the data quality it translates.
Change 4: it's arriving in the UK
For UK retailers the biggest shift is proximity. Google's cross-merchant "Universal Cart" - an AI shopping cart spanning Search, Gemini, YouTube and Gmail, with background deal-finding and back-in-stock alerts - launched with a roster of large retailers and is reported to be rolling out to further markets, with the UK among those to follow. On the ACP side, a major UK sports retailer became, by industry accounts, the first UK retailer to go live on Stripe's agentic commerce suite. Agentic commerce is no longer a US-only story you can watch from a distance; the surfaces your UK customers use are gaining these capabilities now.
That matters because, as covered across this series, UK consumer demand already runs ahead of the European curve while retailer readiness lags. The window between "arriving" and "mainstream" is exactly when the foundational work pays off.
What retailers should actually do now?
The developments are new; the response isn't. If anything, this quarter reinforces the same priorities:
Get the shared foundation right. Every protocol needs clean, complete, structured, machine-readable product data with accurate real-time pricing and availability. This is the one investment that pays off regardless of which camp wins.
Stay interoperable, don't bet on one. Plan to be reachable through UCP and ACP, and let adapter or middleware approaches handle the translation. Avoid a single-ecosystem commitment.
Measure agent traffic now. Attribution won't fully mature for another 18–24 months by most estimates, so the retailers establishing a baseline today will have the data to prove ROI when others are still guessing.
Don't chase the front-end. The in-chat checkout wobble is the cue: build the durable data layer, not the feature of the month.
How VE3 helps
VE3 is a global technology consultancy specialising in data, AI, cloud, and digital transformation. We help retailers get ready for agentic commerce from the foundation up - building the clean, structured, governed product data every protocol depends on, and planning low-integration adapter approaches that keep you reachable across UCP, ACP, and whatever follows, without betting on one or replatforming. Our stance is deliberately vendor-neutral. To make it easy to start, we offer a two-week AI-readiness discovery that shows exactly where you stand against the agentic shift and what to fix first.
Note: the agentic-commerce landscape is changing rapidly; protocol, product, and partner details reflect reports as of July 2026 and should be verified against primary sources.
Want to know whether AI agents can find, trust, and transact with your products? VE3 is offering a two-week AI-readiness discovery - get in touch to book yours.
