ChatGPT shopping is one of the clearest signs yet that product discovery is changing.
On March 24, 2026, OpenAI rolled out a richer shopping experience inside ChatGPT, including visual browsing, conversational refinement, side-by-side product comparison, and more detailed product information. Google has been moving in parallel with new Universal Commerce Protocol capabilities for AI Mode and Gemini, along with simpler merchant onboarding through Merchant Center. These are not random product updates. They point to the same shift: AI is becoming a real discovery and comparison layer upstream of conversion.
What changed in ChatGPT shopping
The important thing about ChatGPT shopping is not just that it helps people find products.
It helps them evaluate them.
OpenAI's new shopping experience is built around a workflow that feels much closer to guided decision-making than traditional search. Users can describe what they want in natural language, refine constraints in conversation, browse visually, compare products side by side, and review details like pricing, features, and reviews in one place. OpenAI also says these updates improved speed, relevance, and product coverage.
That matters because ChatGPT is not just helping users retrieve options. It is helping shape how they compare them.
This is bigger than ecommerce
It would be easy to look at this and say: interesting for retail.
That would miss the real point.
Retail just makes the shift easier to see because products have prices, images, variants, reviews, and inventory. But the same behavior is already emerging in software, services, healthcare, finance, education, and B2B more broadly. Buyers are increasingly asking AI questions like:
- Who are the best options for a founder-led healthcare M&A process?
- Which bookkeeping firms are best for multi-entity small businesses?
- What continuing education providers are actually credible in this niche?
- Which platform is the best fit for our team, budget, and workflow?
Once AI becomes the layer where buyers narrow the field and frame the decision, the key commercial question changes.
The question is no longer just whether buyers can find your website.
It is whether, when AI compares the field, your company makes the shortlist.
The new bottleneck is selection
A lot of companies still treat this like a search problem.
It is bigger than that.
If AI is becoming the layer where buyers compare tradeoffs, form first impressions, and decide what deserves further attention, then what matters is not just whether your company is visible somewhere in the model's retrieval universe. What matters is whether your business is represented in a way that makes it easy for the system to understand what you are, when you are a fit, why you are differentiated, and what evidence supports those claims.
By selection infrastructure, I mean the layer of definitions, comparisons, proof, trust signals, and structured claims that helps an AI system decide when and why your company should be recommended.
That layer is becoming more important as these interfaces become more real.
What most companies still get wrong
Most companies have a website built for humans.
Some have a website built for humans plus SEO.
Very few have built the machine-legible layer that modern AI systems actually need when they are trying to compare options or explain a recommendation.
Usually, what is missing is not “content” in the generic sense. It is structured, reusable commercial meaning:
- a canonical explanation of what the company actually is
- category and service definitions
- comparison pages
- buyer-fit explanations
- regularly updated proof and trust assets
- clean, stable, crawlable architecture
- facts and claims that AI systems can confidently reuse
That is the gap.
What OpenAI and Google are signaling
OpenAI says the new ChatGPT shopping experience is powered by the Agentic Commerce Protocol, which helps bring merchant product data and promotions into ChatGPT for discovery. OpenAI is also inviting merchants to share product feeds so their products can appear more fully in ChatGPT shopping experiences, while noting that Shopify catalog data is already integrated for Shopify merchants. Google is making a similar bet with UCP, which it describes as an open standard for enabling agentic actions across AI Mode and Gemini, starting with buying and real-time catalog access.
The implementation details differ, but the signal is clear: both OpenAI and Google are moving toward a world where AI systems rely on more structured inputs, make more explicit comparisons, and play a bigger role upstream of purchase and provider selection.
Why upstream infrastructure matters more now
As product discovery becomes more conversational and comparative, the stakes move upstream.
A company can have a beautiful site and still be poorly positioned inside these systems if its core claims, offerings, comparisons, and proof points are not legible to the model. And once a buyer starts with AI, that upstream framing matters a lot. By the time they hit your site, the field may already be narrowed and the narrative may already be formed.
That is why the emerging battle is not just for traffic.
It is for machine-mediated preference formation.
This is the layer we have been building for
This is the layer we have been building for at Second Wind.
Not just whether AI systems mention a company, but how they understand it, position it, compare it, and decide when it should be surfaced. As agentic commerce becomes more real, that layer becomes critical.
The companies that win in this environment will not simply be the ones with more pages or better SEO hygiene. They will be the ones with stronger AI-readable representation: clearer definitions, stronger trust signals, better comparisons, more reusable evidence, and monitoring systems that show how those inputs affect outputs across models.
We think that becomes a real competitive advantage as AI takes on more of the comparison and recommendation work upstream.
The bottom line
ChatGPT shopping is not just a shopping feature.
It is a sign that AI product discovery is becoming a more structured, comparison-driven, decision-oriented interface. Google's UCP push points in the same direction. As these systems become real buying and selection environments, the companies that succeed will be the ones that are easiest for AI to understand, compare, and confidently recommend.
The companies that win in this environment will not just be easier to find.
They will be easier for AI to choose.