Book a call →
shopping 2.0
The basket fills without anyone opening a shop. The decisive question is no longer whether customers can find you — but whether the machine still sees you.
ask & done
agentic commerceA year ago the sentence still sounded like science fiction: generative AI would take over online buying. Today it is a plain description of the state of things.
The process has acquired a name — agentic commerce — and works on a strikingly simple principle: the user states a goal, the agent does the rest. You tell an AI: “Order printer cartridge XYZ at the best price, delivery by tomorrow.” Or: “Book me a non-stop flight to London under 600 euros next week, no night flights.” And the agent searches, compares prices, checks availability and return policies, and completes the purchase — without a single shop tab being opened.
The difference from the old e-commerce is fundamental: the user no longer navigates through categories. They state a goal — and delegate.
This is more than a more convenient recommendation. Classic systems showed passive suggestions. The new agents act: they interpret complex intentions, weigh up reviews, reach a decision and trigger the order. McKinsey puts the annual value contribution of generative AI in retail at 240 to 390 billion dollars — a scale that explains why all the big players are suddenly moving at once.
the infrastructure is in place
acp & ucpFor an agent to actually buy, more than a clever language model is needed — it takes a shared language between AI, retailer and payment system. That is exactly what emerged in 2026, practically overnight.
ACP
The Agentic Commerce Protocol, developed jointly by OpenAI and Stripe, enables transactions directly inside ChatGPT — you no longer leave the conversation to buy.
UCP
The Universal Commerce Protocol was presented on 11 January 2026 at the NRF retail show, in a joint keynote by Sundar Pichai and Tobi Lütke. It is open source and is already backed by Etsy, Target, Walmart, Wayfair, Best Buy and Home Depot, along with Visa, Mastercard, Amex, Stripe and Adyen.
What that means
The routes by which an AI retrieves product data, queries prices, checks availability and completes purchases are now standardised. Platforms such as Shopify handle these protocols in the background, so retailers barely notice the technology. Google's Gemini now offers a shopping tool too, taking over selection and purchase within the conversation.
the reality check
harder than it soundsImpressive as this is, an honest look belongs with it, and it tempers the hype. Fully autonomous purchasing across millions of arbitrary retailers is technically demanding. OpenAI initially came up against three hard obstacles: real-time synchronisation of stock levels across countless shops, correct tax collection, and fraud prevention for agent-based transactions.
The consequence is telling: OpenAI shifted focus — away from the complete checkout and towards product discovery. The purchase itself stays with the retailer for now. Agentic commerce is therefore not a finished switch to flip but a process still sorting itself out step by step. That takes nothing away from its force — but it does shift the most urgent question for brands.
geo is the new seo
visibilityWhen an agent steps between the user and the purchase, what it means to be visible changes.
For years the aim was to sit as high as possible in Google's results — the goal was the human click. In the agentic web something else counts: whether an AI includes your products and content in its answer at all.
This new discipline is called generative engine optimisation, or GEO. It extends classic SEO with machine-readable, clearly structured, citable content — so that ChatGPT, Perplexity, Amazon's Rufus, Apple Intelligence and Google AI Overviews understand a brand and reproduce it correctly. In Germany, AI overviews already appear for around a third of all search queries.
Visibility is no longer decided by ranking position but by whether the AI takes you into its answer.
In practice that means a shop that wants to be purchasable by AI agents needs structured product data and clean feeds, documented interfaces, real-time updates on price and availability, schema markup and machine-readable standards. Data quality is the new shopfront. What the agent cannot read cleanly does not exist for it.
brand matters more
takeawayThere is a widespread fear that in a world of automated purchasing the brand becomes irrelevant — the agent only optimises for price and availability, after all. The opposite is true.
AI agents favour brands anchored in the knowledge graphs of language models as trustworthy expert entities. It is not the individual keyword that decides but a brand's authority as an entity — how well a model knows it, places it and passes it on. The same logic sits behind this as behind the classic B2B funnel: wait until the moment of purchase to stand out and you are too late.
Agentic commerce automates the transaction but not the decision about which brand the agent considers in the first place. That decision is made earlier — through data quality, through authority, through a distinctive position. Anyone who thinks they can hide behind the lowest price ends up in a price war. A brand that brings a clear stance and clean data gets recommended.
Why only five per cent of an audience are ready to buy at all, and what that means for the funnel, is covered in detail here: B2B Marketing — Google Ads and LinkedIn Ads combined
Shopping 2.0 is therefore not a threat but an invitation to get your own foundation in order: the data machine-readable, the brand unmistakable, the position unambiguous. The AI handles the purchase — but whether it chooses your brand depends on what you gave it beforehand.
Sources
- 01digitalsprung.de (2026): “Shopify Agentic Storefront — Guide 2026” — ACP (OpenAI/Stripe), UCP launch 11.01.2026 (Pichai and Lütke, NRF), participating retailers and payment providers.
- 02internetwarriors (2026): “Agentic Commerce 2026” — the ask-and-done principle, OpenAI's three obstacles (inventory, tax, fraud), pivot to product discovery, GEO, the 95-5 rule.
- 03norisk Group: “Agentic commerce — how AI is changing online retail” — autonomous purchase steps from analysis to transaction, conversion effects.
- 04rhiem-intermedia (2026): “Agentic Commerce in Shopware” — structured product data, APIs, tokenised payment (Visa Intelligent Commerce), schema and LLM.txt.
- 05plentyOne (2026): “AI and automation in e-commerce — guide 2026” — GEO vs. SEO, McKinsey $240–390bn value contribution, Perplexity, Amazon Rufus, Apple Intelligence.
- 06ZDFheute (2026): “Google tool for online retail — when AI takes over the shopping” — Gemini shopping, convenience and new dependencies.
- 07McKinsey: estimate of the annual value contribution of generative AI in retail; personalisation lifts conversion by up to 20 %.
visible to machines?
Structured data, clean feeds, a brand with a position. We make you findable for people and for agents.
