Morning All,

​
Last October, Amazon boss Andy Jassy more or less read the physical store its last rites. He told analysts that AI would speed up the shift from shops to ecommerce, sketching a future where the customer asks, the machine finds, and the platform delivers, with the shop floor left behind like an old phone box.

He's half right.

And the half he got wrong is the bit worth your attention this morning.

AI is going to do the most damage to boring stores. Now that consumers can compare prices, read the reviews, and have the thing delivered without leaving their sofa...the shop that offers nothing but shelves, a queue, and a teenager who'd rather be anywhere else has very little left to defend,

Here's the part that took me a minute to spot. The exact same thing is quietly happening to brands...and the exact same thing is quietly happening to people.

And that's what we're connecting today.

​
​AI doesn't kill stores, it exposes the dull ones

Start with the shop floor, because it's the clearest version of the pattern.
A recent ICSC and McKinsey report found that 68% of consumers had used at least one AI-enabled shopping tool in the previous three months, and 62% used it to compare brands, models, prices or reviews. The customer walking through your door already knows more than they used to. That raises the bar for what the visit has to deliver.
The operational case for AI in retail is genuinely strong, and worth spelling out. The same research suggests AI, automation and analytics could lift retailers' gross operating profit by up to 20 percentage points, cut out-of-stock's by around 20%, and push conversion up by 15% through real-time in-store data.
Every retailer would bite your hand off if you could deliver that. But notice what they describe.

Better forecasting, better stock, better scheduling. Plumbing, basically.

The thing AI does most ruthlessly is expose laziness and mediocrity.

Tired stores, weak service, dodgy stock counts, the quiet rot that customers used to tolerate because they had no better option. Now they do. Habit won't save a dull store...at least not for long. Location won't. Even a decent range won't, because the customer can see the whole market from their phone before they've put their shoes on.
What's left for the store is the stuff a machine struggles with.

Does this actually suit me?

Is it worth the money?

Can I trust this person's recommendation?

Those questions don't vanish because AI exists. If anything, they get louder.

​
​The floor just got cheap for brands too

Now swap "store" for "brand," and watch the same logic play out.
There's a new tool called Zawa, that turns a rough business idea into a complete brand system in minutes. Logo, colours, typography, the lot, with a "brand memory" feature that keeps every poster and post consistent afterwards.
Then there's the rise of AI user-generated content tools, the ones that spin out scrappy, iPhone-in-the-kitchen style video ads on demand. The promise being sold to DTC brands is that you can test twenty ad angles in an afternoon instead of two a month, and that a solo founder can now match the output that used to need a six-figure studio. That promise may not be strictly true today but the underlying shift is real enough. The floor for looking professional just dropped through the basement. That's wonderful and slightly dangerous in the same breath.

Wonderful, because a one-person business can finally look like it has a marketing department.

Dangerous, because when everyone can look the part, merely looking the part stops meaning anything.

We're already drowning in the result, soulless "AI slop" that scrolls past without leaving a mark.

And here's the tell hiding in both tools. Zawa is brilliant for prototyping, but designers use it to draft and then refine by hand, and bigger businesses outgrow it really quickly. The AI ad tools are pitched for top-of-funnel testing, not the long, patient work of building a brand people love. The market is splitting cleanly into craft and depth vs speed and volume.

​
​Ulta shows you how it's supposed to work

If you want proof of the grown-up version, look no further than Ulta Beauty.
Net sales rose 11.1% to $3.16 billion, with online L4L sales growing in the mid-teens while brick and mortar stores grew low single digits. The digital side is clearly motoring. And AI is doing real work underneath it.

A shopping assistant called Ulta AI built on Google's Gemini, plus agentic commerce inside Google's AI Mode, so a shopper can search, compare and buy Ulta straight through Google's chat.
I've consistently argued Google is best placed to win the AI wars because it's quietly powering everyone else's tools, and here it is again, sitting as the engine powering Ulta's assistant.
What did Ulta do with all that capability?

They handed the commodity jobs to the machines, the search, the comparison, the refill reminders pulled from 47 million loyalty members' worth of data. Then they spent the freed-up energy on the things people actually fall for.

Exclusive brands. Curation. A "highly experiential" flagship coming to Times Square in 2027.
They didn't use AI to cut their way to mediocrity. They used it to sharpen the bits that make Ulta feel like Ulta.

That's the whole game.

​
​So what's the moat now? Taste.

​
Cast your mind back a couple editions when Andon Labs handed an AI agent called Luna a budget, a corporate card and a real shop to run in San Francisco. Remember? It could hire people, source products, make decisions.

Here's the update...it lost money hand over fist, roughly $14,300 a month in costs against $6,000 to $8,000 in revenue, and it didn't reliably tell people it was an AI when hiring them. A shop, it turns out, is a thousand small human judgements a day, and a machine left alone made a mess of them.
On the other hand, western wear brand Tecovas uses AI in its "Boot Runner" programme to fetch a specific pair of boots to the sales floor in about 85 seconds. The machine does the running about. The human stays beside a nervous first-time buyer through a purchase that feels unfamiliar. Their tech chief reckons AI gave them a better human connection in store, not a thinner one.

One version gutted the human. The other amplified them. That's the moat.

Taste, judgement, experience, and the innately human knack of getting other people to connect and care about a brand.

Those are the things AI can't cheaply or easily replicate.

There's the catch though, and it's a very important one. The moat only holds if the human is genuinely exceptional. A mediocre operator armed with AI just produces more mediocrity faster.

More boring posters. More boring videos. At scale.

Stores, brands, people now have to know and be crystal clear about what they are for. What they cannot afford to be, is vague, and they definitely cannot afford to be mediocre.
​

​
​If you remember nothing else: When everyone can look polished, polish becomes the floor, not the edge. For the exceptional human who learns to drive the machines, and consequently beats both the human too proud to use them and the machine with no taste of its own...this is a huge opportunity. So this week, be exceptional. Write down in one sentence what your store, brand or you actually stand for. The single thing that sets you apart from both your colleagues and an algorithm. Then step by step hand over every task that isn't that sentence to an AI tool and reserve your judgement for where it truly counts.

See you next week.