Something quietly significant happened in late 2025. OpenAI launched instant checkout features that let AI recommend products and complete purchases directly within the conversation. No click to a website. No landing page. No traditional funnel.
AI platforms are projected to account for about 1.5% of US retail e-commerce sales in 2026, nearly four times the 2025 figure. That might sound small. So did social commerce five years ago.
What Agentic Commerce Actually Means
Agentic commerce is the idea that AI systems don't just answer questions about products. They research, compare, recommend, and complete the purchase on behalf of the user. The consumer says "I need a moisturiser for dry, sensitive skin under 60 euros" and the AI handles the rest.
This is fundamentally different from how search works. In a Google Shopping query, the consumer sees a list and chooses. In agentic commerce, the AI chooses for them, or at least narrows the field to one or two options with a strong recommendation.
The implications for DTC brands are significant.
Your Product Data Becomes Your Storefront
When an AI agent is deciding whether to recommend your product, it's not looking at your beautiful website design or your Instagram aesthetic. It's reading your product data. Descriptions, ingredients, specifications, reviews, pricing, availability.
If your Shopify product descriptions are thin ("Luxurious moisturiser for all skin types"), the AI has nothing to work with. If your competitor's descriptions are detailed ("Ceramide-based moisturiser for dry and sensitive skin, fragrance-free, dermatologist-tested, 50ml"), the AI will recommend them instead.
This is SEO for LLMs. The difference is that traditional SEO optimises for ranking. LLM optimisation determines whether your product exists in the AI's consideration set at all.
What Brands Should Be Doing Now
Structured product data. Every product should have detailed, factual, attribute-rich descriptions. Not marketing copy. Data. Ingredients, use cases, skin types, price per unit, certifications. The more structured and specific, the more likely an AI can match it to a consumer's query.
Reviews and social proof. AI systems weight third-party validation heavily. A product with 200 reviews averaging 4.5 stars is going to get recommended over one with 12 reviews, regardless of the marketing copy.
Schema markup. Structured data on your website helps AI systems parse your product information. If you're not using JSON-LD product schema on every product page, you're invisible to half the systems that will drive discovery in the next two years.
Price competitiveness. When an AI agent is comparing options for a consumer, price is a first-order filter. This doesn't mean racing to the bottom. It means being clearly positioned within your segment and making the value proposition explicit in the data.
The Bigger Picture
The channels that drive e-commerce acquisition have shifted roughly every five years. Search, then social, then influencer, then TikTok. AI-driven commerce is the next one. It's early, and the infrastructure is still forming. But the brands that start optimising for it now will have a meaningful head start when the volume arrives.
The question is the same one it's always been: is your brand discoverable where your customers are actually looking? Increasingly, they're not looking at all. They're asking an AI to look for them.