When I wrote in January about the move from operations to intelligence, I framed 2026 as the year commerce teams would shift from reacting to issues toward proactively shaping outcomes. Four months in, that direction is holding. But the conversations I’ve been having with merchants since then have sharpened my view of how this shift plays out in the field, and where the gap between vendor narrative and merchant reality is widest.
This post is a field report. It’s what I’m hearing across discovery calls, demos and customer reviews. Some of it confirms the strategy. Some of it has surprised me. All of it should be useful if you’re a merchant trying to make sense of an unusually loud AI commerce market.
What I’m hearing most often
Three patterns show up in almost every conversation.
Merchants are tired
Most teams I talk to have already been pitched on AI personalization, AI search, AI chat, AI recommendations and now AI agents. Many have tried something. A meaningful number have been burned by it. The first question is rarely “can you do AI?” It’s closer to “how is this different from the thing that didn’t work for us last time?” The merchants who once asked about features now ask about pricing models, attribution and exit terms.
The buyer has changed
When I joined Webscale, the typical conversation started with an infrastructure or engineering leader. Today, more than half of my AI conversations begin with marketing, ecommerce or digital leaders who control their own budgets. They’re buying outcomes, not platforms. That changes the language, the proof points and the cycle. It also raises the bar on how quickly we have to show value, because these buyers are accountable to revenue numbers that move every week.
The protocol conversation has gotten real
Five months ago, when I mentioned the Unified Commerce Protocol or agentic commerce, the response was usually polite curiosity. Today, merchants are asking direct questions about how AI agents will interact with their catalogs, whether they need to expose product data differently and what happens to attribution when a buying decision is made inside an AI assistant they don’t own. That’s a substantive shift, and it’s happening fast.
Where the conversation stalls
A few things I see across the market that consistently don’t land.
Generic AI chat as a feature checkbox
Merchants have seen enough scripted bots and rules-based personalization engines to be skeptical of anything that looks like a chat widget bolted onto an existing site. The interest is real, but the bar is now whether the system understands their catalog, their customers and their business context. That’s the difference between generic AI chat and a real AI Shopping Assistant. Surface-level demos don’t survive a second meeting.
ROI claims without attribution clarity
Slides promising a 30% conversion lift get nods and no follow-up. The merchants who are serious want to see how you prove the AI drove the sale, not the customer who would have bought anyway. This is one of the most consistent objections I hear, and it’s fair. Vendors who can’t answer it lose the deal.
Enterprise stacks pretending to be agile
Larger personalization and CDP vendors are repositioning aggressively, but for a lean ecommerce team running on Magento or Shopware with one or two marketers, the cost, complexity and time-to-value just don’t work. The merchants moving fastest are the ones who don’t already have a CDP and aren’t interested in a 12-month implementation.
Where the conversation clicks
The opposite is also true. A few things consistently land hard.
Live, first-party data over assumed accuracy
When a merchant sees their own session data flowing into a profile in real time, instead of waiting weeks for cold-start models to learn their customers, the conversation changes. Existing Webscale hosting customers feel this most directly, because their session data is already inside our platform. That head start is something hyperscaler-plus-bolt-ons can’t replicate.
Usage-based pricing as a trust mechanism
For merchants who’ve been burned before, a predictable per-conversation model removes the fear of paying for a platform that might not work. It reframes the buying decision from “do I believe the pitch” to “let’s see what it earns.”
Precision over breadth in regulated and high-consideration categories
In firearms, regulated goods, B2B parts catalogs and high-AOV considered purchases, hallucination isn’t a feature flaw. It’s a liability. The merchants in those verticals want to see how the system constrains itself, where the human stays in the loop and how accuracy is enforced at the catalog level. When that question gets answered well, the conversation accelerates.
What this means for the rest of 2026
The merchants moving fastest right now share a profile. They’re lean ecommerce teams without a deeply entrenched CDP. They’ve likely been let down by a past AI investment. Their buying decision is owned by marketing, not IT. And they care more about provable revenue impact than about the size of the model behind the scenes.
That last point matters most. The AI commerce market is going to keep being loud, and the temptation to chase the model of the week will be constant. The merchants who win in the next twelve months won’t be the ones with the most sophisticated AI. They’ll be the ones whose infrastructure can feed AI live data, act on its output in milliseconds and prove the result.
That’s the work we’re focused on. It’s the conversation I’d like to keep having.
Compare notes with me
If you’re evaluating AI commerce vendors right now, or if you’ve tried something that didn’t work, I’d like to hear what you’re seeing. The patterns in this post came from listening, and I learned something useful from these conversations.
If you’re an agency bringing AI conversations to your merchants, I’d like to compare notes on that too. Some of our best product feedback comes from implementation teams who see what actually works in the field. Our Partner Program is built around that kind of collaboration, and there’s real commercial upside for agencies who get in early on the AI motion.
Get in touch. I read every note.
