That line hits because it’s not a philosophy. It’s logistics.
Harley Finkelstein (Shopify president) was talking about the next interface layer for commerce—AI shopping inside chat—when he said: “It’s not like where you can say you don’t do social commerce. You can’t really say you don’t do AI.” (Source: “How AI Shopping Could Turn Fashion Advertising on Its Head,” Vogue Business, Amy O’Brien, January 14, 2026.)
If you’re a brand, a retailer, a service business, a creator, or a local operator, the sentence is basically a forecast disguised as a quote: AI becomes a distribution channel the way “the web” and “mobile” became distribution channels. And distribution channels don’t ask permission.
Why This Isn’t Optional
Most “AI talk” still sounds like strategy. This one sounds like gravity.
When an interface becomes the place people start—and increasingly finish—their journey, opting out isn’t a stance. It’s simply losing visibility. Years ago, businesses could say, “We don’t do websites,” then “We don’t do mobile,” then “We don’t do social.” Those were all survivable for a while… until they weren’t.
AI is headed toward the same inevitability, except faster, because it compresses steps: discovery becomes a conversation, comparison becomes a summary, decision becomes a recommendation, and purchase becomes a single integrated action.
Once that pattern becomes normal, “I don’t do AI” is like saying “I don’t do roads” while selling products that require delivery.
The Real Shift: From Clicks to Recommendations
The internet trained brands to fight for attention. AI will train brands to fight for being selected.
Classic digital marketing is dominated by auctions: you pay to show up. But in an AI-mediated experience, the user is explicitly asking for the best option for me—not the loudest option, not the most optimized headline, not the brand with the biggest retargeting budget.
That changes the game from “How do I get in front of the customer?” to “How do I become the recommendation?”
And that forces a different kind of discipline. AI systems don’t just read your ads. They read your entire footprint: your product pages, your FAQs, your returns policy, your shipping speed, your reviews, your reputation on forums, your consistency across channels, and the coherence of your brand story.
In other words: the machine is building an opinion of you from the totality of what you’ve put into the world.
Advertising Doesn’t Die — It Mutates
The fear response is: “AI will kill advertising.”
The more accurate response is: AI will move advertising into the conversation itself—and it will reward brands that already earned trust.
The Vogue Business reporting points to emerging models like in-chat “direct offers,” sponsored follow-ups, and branded agents that answer questions in a brand’s voice. The monetization will arrive because the incentives are too large for it not to. But the user’s tolerance for interruption will be much lower inside a conversation than inside a feed.
So the ads that win won’t feel like ads. They’ll feel like a helpful offer at the right moment, a credible option worth considering, and a brand that belongs in the shortlist.
The channel will still be paid in many cases—but the content has to behave like relevance, not like persuasion.
“Merit-Based Retail” Is Real, But Don’t Misread It
One of the most important ideas in the article is the suggestion that AI shopping could become more “merit-based” than the current pay-to-play environment—because an AI agent’s promise is that it’s acting on behalf of the user.
That’s directionally true. But here’s the devil’s advocate view: merit doesn’t mean fairness. Merit means signals.
AI will rank what it can interpret. Brands that produce clearer signals will look “better,” even if a quieter brand has higher craftsmanship. The machine can’t reward what it can’t see.
So the opportunity is huge for smaller brands—but only if they make themselves legible: legible to the machine (structured, descriptive, consistent), and legible to the human (trustworthy, coherent, real).
What “Doing AI” Actually Means for a Brand
This is where people get it wrong. They hear “AI” and think “we should post more AI content” or “we should build a chatbot.”
No. The first wave of “doing AI” is unglamorous:
- Make your catalog machine-readable: longer descriptions, specific use cases, materials, fit, care, comparisons, and the boring details that remove ambiguity.
- Build the FAQ your customers actually ask for: operational FAQs, not marketing FAQs.
- Upgrade imagery for interpretation, not just aesthetics: AI is increasingly multimodal; your images are inputs for understanding.
- Treat reviews as a core asset, not social proof: reviews aren’t just persuasion; they’re training data for trust.
- Fix the last-mile trust points: shipping reliability, customer service responsiveness, returns clarity.
- Make your brand narrative coherent across the web: fragmentation is punished; clarity is rewarded.
- Prepare for in-chat commerce pathways: reduce friction so purchases complete cleanly.
That’s “doing AI.” It’s not a campaign. It’s infrastructure.
The Quote Is a Warning — And a Gift
Let’s treat Finkelstein’s sentence like a sign on the highway: you can keep driving as you are, but the exits are changing.
The winners won’t be the brands who scream “AI!” the loudest. The winners will be the brands who quietly become easier to understand, easier to trust, easier to buy from, and easier to recommend.
That’s why the quote matters. It’s not hype. It’s a recognition that AI is becoming a default interface. And default interfaces quietly decide who gets discovered.
The Practical Bottom Line
If you run a brand and you’re reading this, your next move is not “launch an AI initiative.”
Your next move is to become the kind of brand an AI would confidently recommend—because the evidence is clear, the story is coherent, and the customer experience backs up the promise.
You can’t really say you don’t do AI.
Not because you must love it.
Because your customers will.
And they’ll bring the AI with them.