The biggest players in tech have been locked in a race to own the future of shopping. In the last few years alone, OpenAI rolled out shopping features, pulled back, and then reintroduced them.
Perplexity launched a buy button. Amazon built “Buy for Me,” much to the chagrin of retailers who did not opt in. Google redesigned shopping around AI answers. Even though everyone is constantly pivoting, nobody has a clear win.
It’s because they’re all focused on the wrong problem.
The conversation about agentic commerce has almost entirely centered on consumer discovery and checkout. Things like how AI helps someone find a gift, compare mattresses, or click “buy” through a chatbot. While that’s a reasonable place to look, it’s also the hardest possible place to start, and potentially the last place AI will deliver value.
Meanwhile, the boring, unglamorous, enormous world of B2B procurement sits largely unexamined. That’s where agents have the biggest potential to break through first.
When AI rewrites a checkout flow, it doesn’t inherit a blank slate. It inherits years of painstaking optimization. Brands have spent enormous resources perfecting the moment between “I want this” and “I bought this.” Every upsell, loyalty touchpoint, and data capture moment exists because brands worked out what moves a buyer forward.
An AI agent makes that optimization irrelevant. It doesn’t care about carefully designed product pages, email capture, or recommendation engines. The buyer gets an AI answer based on simple data in the product catalog, with marginally less friction on an already functional experience.
Early data backs this up. Walmart reported a 66% drop in conversions when agents intermediate the buying experience. Amazon’s “Buy for Me” triggered backlash from retailers because they optimized for buyers at the explicit expense of sellers.
Some argue that consumers don’t trust AI to shop for them. At the same time, people hand their credit card to Temu without blinking, despite FBI warnings of data risks. Convenience and value drive adoption, not trust. The problem is that nobody has built a consumer AI commerce experience that clearly beats the existing one.
B2B procurement is a different animal
Let’s contrast that experience against B2B buying, which is a completely different process built around complexity, not convenience.
Consider what a mid-market manufacturer navigates to buy something as routine as industrial fasteners. They might use approved vendor lists, or take advantage of negotiated contract pricing that differs by account.
They almost certainly have compatibility requirements, complicated approval workflows that vary by order size, or a purchase order generation process that touches an ERP. Meanwhile, somewhere buried in all of the complexity is the specific screw they need, and they probably need a sales rep to get it.
Here’s the thing: most of what those sales reps do isn’t strategic judgment. They’re applying institutional knowledge to a decision tree that, if anyone had bothered to encode the rules properly, would run itself. That is exactly where AI agents can give sales reps superpowers.
Why B2B may lead consumer for the first time
Throughout tech history, consumer products achieved massive scale, then enterprise versions followed. For example, social media was Facebook before it was Slack. Agentic commerce may flip that model because B2B organizations have better data.
But that doesn’t always mean cleaner data. Most B2B product catalogs are a disaster of PDFs, legacy ERP exports, and specs living in salespeople’s heads. But they have massive potential to be more structured, rule-bound, and useful to a machine.
While a consumer brand’s product data is images, a description, and a star rating, a B2B distributor’s catalog is a relational system of SKUs, compatibility matrices, account-specific pricing layers, and contractual constraints. While the consumer data may look richer, the B2B data is more actionable.
The scale of the opportunity makes this even more compelling. B2B e-commerce globally is roughly five to six times the size of B2C by transaction volume. Eliminating even a fraction of procurement friction will restructure the entire category.
Making data readable by machines
None of this works without confronting the real problem underneath both B2B and consumer AI commerce: product data isn’t built for machines. Commerce infrastructure has been optimized for presentation to humans.
But an AI agent doesn’t interpret how well a website is optimized for conversions. It needs to know which products are compatible, what’s in stock, and what the price is for this specific buyer under this specific contract at this specific moment. If that information isn’t explicit and structured, the agent doesn’t ask a clarifying question. It moves on.
The companies that solve this in B2B first will have a significant head start, because B2B product relationships are already more explicit. Compatibility isn’t a matter of taste, and pricing isn’t based on “vibes.” The rules are all there; they just need to be encoded.
Consumer commerce will get there too. But it will take longer, cost more, and require breaking things that currently work before building something better. B2B procurement doesn’t have that problem. Most of it’s already broken. AI’s biggest challenge is a fax machine and a sales rep eating lunch.
That bar to fix B2B commerce is lower than anyone seems to realize. And clearing it will be worth a lot more than fixing checkout.
We’ve featured the best AI tool.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
https://cdn.mos.cms.futurecdn.net/vUGGQXBvMLxK65oJegUwgk-2000-80.jpg
Source link
