What is agentic commerce? How AI agents will shop for fashion customers

Agentic commerce is shopping in which an AI agent acts for the customer: it interprets a request, searches and compares products, and can complete the purchase within limits the shopper sets. For fashion, this means a growing share of discovery and some checkout may happen inside assistants such as ChatGPT, Gemini or Perplexity rather than on a brand's own website.

What is agentic commerce, and how is it different from a chatbot?

A conventional shopping chatbot answers questions on one retailer's site. An agent goes further: it plans multi-step tasks, calls external systems and takes actions. McKinsey's October 2025 report defines agentic commerce as shopping powered by AI agents acting on our behalf, with agents that anticipate needs, navigate options, negotiate and execute transactions.

The same report distinguishes three interaction models, which are useful for fashion teams thinking about where their products will be seen and bought.

How big could agentic commerce become?

Forecasts are early and should be read as scenarios. McKinsey estimates that by 2030 agentic commerce could orchestrate 900 billion to 1 trillion US dollars of US B2C retail revenue, and 3 to 5 trillion dollars globally, for goods only. The report also notes that half of consumers already use AI when searching the internet.

Behaviour is shifting faster than transactions. Most fashion purchases today still end on a retailer's website or app, even when an assistant started the journey. That distinction matters for planning: preparing to be found by agents is urgent, while preparing to be bought from inside an agent depends on how platforms settle their checkout models.

Read also

How is AI search changing the way shoppers find fashion?

Which platforms are building agentic shopping today?

Several large platforms launched agentic shopping features between late 2025 and 2026, and their approaches are still changing.

• OpenAI: launched Instant Checkout in ChatGPT on 29 September 2025, starting with US Etsy sellers, built on the Agentic Commerce Protocol co-developed with Stripe. The merchant remained merchant of record. In March 2026, CNBC reported that OpenAI was moving away from Instant Checkout towards retailer-run apps inside ChatGPT, with purchases completed on the retailer's own site.
• Google: announced the Universal Commerce Protocol in January 2026, co-developed with Shopify, Etsy, Wayfair, Target and Walmart and endorsed by more than 20 companies including Zalando, Macy's Inc., Visa and Mastercard. It powers checkout on eligible listings in AI Mode and the Gemini app for eligible US retailers, who remain seller of record.
• Retailers' own assistants: Zalando reported in March 2026 that its assistant had six million users, four times as many as a year earlier, and that it was preparing for purchases through chatbots via Google's protocol.

What does an AI agent need from a fashion brand?

Agents work from structured data, not from campaign imagery or brand storytelling. In apparel, the attributes that matter most are the ones that cause returns and failed searches when they are wrong: size and size system, colour, material, fit, gender and age group, variant grouping, price, and live availability per size.

McKinsey's recommendations to merchants point in the same direction: make product data agent-readable, build modular APIs instead of relying on a single exclusive AI partnership, and design an agent experience alongside the customer experience. For most fashion businesses, the first step is not a new AI project but fixing product information management and stock feeds.

What are the risks for fashion retailers?

The opportunities come with real trade-offs that deserve sober assessment.

1. Disintermediation: if the agent owns the conversation, the brand may lose direct contact, data and the chance to cross-sell. McKinsey warns that agents could become the new gatekeepers of commerce.
2. Ranking opacity: it is not always clear why an agent recommends one product over another, and eligibility for a platform does not guarantee display.
3. Errors and liability: agents can misstate sizes, prices or materials. McKinsey flags brand liability and regulatory scrutiny as risks.
4. Retail media pressure: if shoppers bypass search results and product pages, advertising income tied to those pages may weaken.
5. Platform volatility: OpenAI's change of course on checkout within six months shows that commercial models are not settled.

How should fashion brands prepare in 2026?

A realistic plan focuses on no-regret moves: complete and consistent product attributes, accurate size-level stock, product feeds to the major assistants where programmes exist, clear returns and delivery information in machine-readable form, and analytics that separate AI referral traffic from other channels. Commercial decisions, such as whether to enable in-assistant checkout or build an app inside an assistant, can follow once the economics and terms are clearer.

Read also

Launching an AI shopping assistant: a 20-point checklist for fashion e-commerce

Will AI agents replace fashion e-commerce sites?

Not in the near term. Platforms themselves are still sending many shoppers to merchant sites to complete purchases, and fashion relies heavily on visual inspiration, fit confidence and brand experience that assistants only partly replicate. The more likely outcome is a mixed model in which agents handle more discovery and comparison, while brand sites and apps remain important for checkout, loyalty and service.

For fashion executives, the most useful framing is a staged one. In the first stage, assistants act as advisers that send shoppers to brand and retailer sites; this is already happening and mainly rewards good product data and clear policies. In the second stage, assistants complete purchases within platforms, with the retailer still fulfilling and serving the customer; this is being tested, and terms are still changing. In a third stage, personal agents may act with more autonomy, for example reordering basics or watching for a specific item in a given size and price range. Each stage raises the importance of accurate size-level stock and transparent return conditions, because the customer may never see the product page before buying. Planning for the first stage now, while monitoring the second and third, avoids both over-investment and being left out of a channel that is growing quickly.