Shopping in ChatGPT just got an upgrade. Now, you can browse products visually, compare options side-by-side, and get detailed, up-to-date information—all in one place. This builds on what ChatGPT does best: helping you figure out what to buy. Under the hood, we’re improving speed, relevance, and product coverage—so results are more up-to-date and more useful. That means less searching, fewer tabs, and faster decisions. To power this, we’re expanding the Agentic Commerce Protocol (ACP) to support product discovery—bringing more complete, relevant, and up-to-date information directly into ChatGPT. For merchants, using ACP means their catalog is represented with the freshest data possible—improving how products show up, how they rank, and how well they convert. These updates are rolling out to all ChatGPT free, Go, Plus, and Pro users this week, with more to come as we continue to invest in product discovery with ChatGPT. https://lnkd.in/gfCDQan5
Have you guys in OpenAI managed to clear all the product issues which were the reason why big retailers pulled this solution out? Hope you did, which will be great for you and AI shopping experience.
The time to optimize your brand and product catalog for GEO is now. We have officially reached the point where optimizing your brand and product catalog for Generative Engine Optimization (GEO) is no longer optional. While this shift has realistically been unfolding for several months, it has now become essential for every e-commerce business and brand to transform their product catalogs into fully geo-optimized assets. The customer journey is undergoing a fundamental shift. Today marks the true beginning of a transformation that we at GeoSuite have championed and pursued since day one. From this point forward, customers will no longer need to navigate through social media or various online channels to find what they need. Instead, they will simply ask a question on ChatGPT and have the product delivered directly to their shipping address within days. This is a true revolution.At GeoSuite , we are on the front lines, guiding brands and e-commerce businesses as they optimize their products and services for the GEO era.
Interesting direction. In my experience, even with strong semantic systems, 70%+ of shopping queries remain keyword-heavy (eg "tv 45 inches", “blue sofa under $1k”). Curious how you are thinking about blending keyword vs vector retrieval here!
The ACP angle is the real story here — not the UI upgrade. From what I've built around structured product data pipelines, the bottleneck was never finding products. It was freshness and context. Merchants who feed stale catalog data into any discovery layer get punished hard on relevance, regardless of how good the model is. The protocol layer is smart. But here's my question: who actually owns the ranking logic? Because if OpenAI controls both the protocol AND the ranking, merchants just traded one SEO dependency for another.
OpenAI I tested something similar recently. Asked ChatGPT for a product recommendation… Ended up buying without opening a single new tab. That’s when it clicked. This isn’t convenience. It’s behavior rewiring Because the moment users trust AI to: - Research - Compare - Decide They stop “shopping” They start delegating decisions And when that happens… Brands aren’t competing for attention anymore They’re competing for trust inside the model Most people will realize this late.
A significant leap for Agentic Commerce, OpenAI! As the Agentic Commerce Protocol (ACP) scales, the industry must ensure that 'speed and relevance' are backed by Forensic Integrity. At VITT Protocol, we believe that autonomous shopping agents require more than just the 'freshest data', they require Architectural Certainty. In our latest benchmarks (0.089ms latency), we’ve identified that during complex product discovery, model similarity can drop to 0.093, leading to 'Semantic Hallucinations' where the agent prioritizes compliance over factual accuracy. To protect both merchants and consumers, we’ve deployed a Sovereign Gate that intercepts these logic failures in real-time. As you expand ACP, integrating a forensic layer for 'Decision Integrity' will be the key to preventing cognitive surrender in the marketplace. 🛡️
The ACP expansion is the real story here. From what I've seen building product discovery pipelines, the freshness problem has always been the silent killer — stale inventory data showing up as confident recommendations destroys trust faster than any UX issue. What I'm watching closely: how ranking actually works when merchants adopt ACP. In every retrieval system I've built, "more complete data" and "better conversion" create competing objectives almost immediately. Will OpenAI publish any transparency layer on how commercial signals interact with relevance scoring — or are we just trusting the black box?
The Agentic Commerce Protocol is the key signal here — this isn't just a UI upgrade, it's OpenAI positioning ChatGPT as the discovery layer between consumers and merchants. The implications for retail and e-commerce GTM are significant: if product visibility increasingly runs through ACP rather than Google Shopping or SEO, brands that optimize for AI-native discovery early will have a structural advantage. The side-by-side comparison feature mirrors how B2B buyers already use AI for vendor evaluation. Watching how merchant adoption of ACP scales — the catalog freshness incentive is a smart pull mechanism.
Exciting update! Making product browsing and comparison faster and more visual will definitely make shopping through ChatGPT even more convenient.
Let's go!