EMPIRE OF AI: Welcome to the Real World
AI and Commerce: There is no spoon
“Welcome to the real world.”
— Morpheus, The Matrix
Let’s get this out of the way up front: I’m a skeptic. A well-meaning, slightly-exhausted, time-crunched, commerce-loving skeptic who’s been around long enough to remember when “personalization” meant someone spelling your name right on a marketing email.Does the fact that we still get this wrong tell us something.
And yet… here I am. Writing 6,000 words about artificial intelligence and its slow, relentless invasion of commerce, like Agent Smith, but wearing a branded hoodie and holding a DTC skincare subscription. That’s my vibe and I am sticking to it. I am trying to research the topic in a way that has more meaning for me personally. Enter Karen Hao’s Empire oif AI. A really great read, well listen. Audiobook for me. The book starts just over 10 years ago. AI was well before that but AI as we view and know it today, started around then. The Sam Altman version, let’s call it. Structure and infrastructure give a good indication of where true power and true conflict lie. How do we make this better for people and governments were early, altruistic ideals. Today, there is so much infrastructure around that anywhere there is data, there remains opportunity. But for who?
Because even I can admit: something fundamental has changed. Not in the rah-rah, Silicon-Valley-LinkedIn-influencer sort of way. But in the quieter, more unsettling sense that the interface of commerce has been hijacked. Not by a hacker. Not by a brand. But by the algorithm. I was recently told of the white space Tobi encouraged the Shopify network to think about. White Space seems less hopeful to me than its aspirational design. I am trying to make sense of it all. If not for anyone, but for myself firstly.
And to make sense of it, we need to take the red pill.
1. The Origins: “You take the blue pill...”
“You take the blue pill—the story ends, you wake up in your bed and believe whatever you want to believe.”
— Morpheus, The Matrix
The early history of AI in commerce is the equivalent of that scene in every prequel where the villain is still just a misunderstood intern. We were dealing with “expert systems,” not existential threats. Rules-based engines. Linear logic. Nothing that questioned the very fabric of capitalism, just tools to help us keep a few more pallets off the clearance aisle.
These were the blue pills of ecommerce. Innocent. Boring. Mostly Excel in disguise. The very notion of being sold to in this way is not new to us.
1.1 Inventory: The Original AI Daydream
Back in the 60s and 70s, companies like Walmart were fiddling with just-in-time inventory and point-of-sale data. This wasn’t AI, it was glorified counting. But it sowed the seeds.
Fast forward a few decades: algorithms start setting prices based on demand forecasts. Amazon begins adjusting product listings based on click rates. It’s not AI, yet. But the Matrix is forming. Prime Day just ran a 4 day deals extravaganza (bluegh) and Rufus played its part. But what if I said that 20% of our waking lives are dictated by the fabric of AI.
1.2 Recommendation Systems: “People who bought this also bought…”
Suddenly, AI becomes visible. It’s not just churning in the background it’s suggesting things. Poorly, at first. Like when Netflix kept recommending Paul Blart: Mall Cop because you watched one documentary on shopping malls. Seriously, how do some movies get made. Hopefully mine will too, in that vein.
But this was the first sniff of what was coming: consumer-facing machine logic. “If you like this, try this.” What seemed like harmless utility would evolve into manipulative precision.
1.3 Fraud Detection: The Cops of the Codebase
AI’s most successful early role? Catching fraudsters. Pattern recognition saved banks and merchants millions. Unfortunately, it also introduced the world to algorithmic bias flagging transactions based on behaviors it barely understood, from people it barely cared to. That said, Frank Abagnale told the MRC not so long ago that it was easier to perpetrate fraud now than it was during his years as a pilot, or doctor or was it a lawyer?
We didn’t know it then, but these early applications were gateway drugs. We thought we were building tools. We were actually building agents.
2. The Inception: “The Matrix is everywhere…”
“The Matrix is everywhere. It is all around us. Even now, in this very room.”
— Morpheus, The Matrix
If the 2010s were about mobile eating the world, then the 2020s are about AI swallowing the remains. Just ask Ben Evans.
We’re now in what Scot Wingo, a man who’s seen more commerce trends than I’ve had coffees this week, calls the "agentic" era. Think less Siri, more Skynet... but with a Shopify plugin.
This isn’t AI as backend tool anymore. It’s AI as lead actor. His substack RetailGentic is one of the more interesting on the subject matter with a great perspective too.
2.1 Agentic AI: Your New Personal Shopper (With No Commission Incentive)
Wingo defines agentic AI as systems that operate with autonomy, set goals, interact with environments, and execute tasks.
Let’s spell that out:
It doesn’t wait for your prompt.
It doesn’t need your oversight.
It doesn’t ask permission.
It does your shopping, finds your supplier, negotiates your discount, and emails the warehouse to prep the order. All while you’re watching a Netflix show that it recommended.
This is no longer your tool. It’s your proxy. Your commercial doppelgänger.Your digital twin maybe. Fu*ckin digital twins - matrix number 2.
2.2 From Backend to Interface
While most people didn’t notice the moment AI moved from logistics to the storefront. One day we had dropdown filters, the next we had GPT-powered concierge chatbots acting like they knew us better than our own families. Let me be clear at this point though - we never asked for this.
Never, as consumers - the way this works in the long term is not if it is an option it is if it is the only way. We respond better to no choices as opposed to overload these days.
AI agents are everywhere now:
ChatGPT + Klarna partnership? That’s an agent.
Google’s “Circle to Search”? That’s a halfway house.
Perplexity’s shopping results? An agent pretending it’s still doing research. Soon to be fed by Feedonomics.
For everything else there’s … I won’t.
We are moving fast from clicks to delegation. And commerce at least the parts not asleep at the wheel is responding. Do we have time to think about it? To respond and change?
3. The Expansion: “What is real?”
“What is real? How do you define real?”
— Morpheus, The Matrix
Commerce used to be about catalogues, stores, websites. Then it became mobile. Now it’s something else. Something... less visible. More synthetic. And less human. A recent look at the stunning Oh Polly PLPs feels like a hybrid in motion. A look at what it could be. Thanks Ger.
We don’t see AI running the show, because it doesn’t show itself. But it’s writing the scripts. Literally and figuratively. Those days are not too far in front of us though. Big bets backed by big investments means we will be facing into our agentic indian summer before long. The long term costs in investment will be paltry to a 100 year company like Shopify in the future. To quote the book and Sam Altman - we want to get more revenue so we can do more research, not acting in the reverse like most.
3.1 The New Commerce Stack (Now with 80% More Invisible Hands)
POSITIVE IMPACTS THE ARGUMENTS FOR
Scale without burnout: AI does what humans can’t or won’twithout coffee breaks or unions. I grew up in a union family so I know their value. But AI is now part of all of our routines. We have been convinced of its place and its purpose. Hard to deny it. We may not yet even be mid level users but we need to become experts. Dealing in ambiguity and certainty in the same context is something we need to do.
Personalization++: No more "Hi [FirstName]" emails. Now it’s pricing based on your lifetime value and what your friends might spend. And any other data that deep learning modals are willing to share with each other, which is pretty much everything.
Intelligence, not just data: AI spots patterns in returns, basket size, conversion cycles, TikTok trends, and your dog's birthday.
Always-on commerce: Chatbots that don’t sleep. Dynamic pricing that refreshes hourly. Campaigns that test themselves. But don’t tell the state of New York.
NEGATIVE IMPACTS
Job redundancy: If your job is repeatable, it’s deletable. Sorry Karen from Customer Service. Sorry all of us. Jobs numbers will be impacted, more than we think or want to get our heads around. To be valuable to nations, there needs to be economic value. This is where some conflicts and contradictions appear.
Bias at scale: Garbage in, garbage amplified. From hiring filters to product recommendations. But once it learns, itself, won’t it create less garbage and reject existing garbage?
Loss of transparency: You don’t know why it did what it did. Neither does your dev team. Transparency leads to corruption allegedly and wars we dont want to fight.
Market collapse (in parts): Legacy brands that can’t keep up with real-time adaptation? Flattened by the AI curve. Legacy technology that can’t keep up will die. New companies who think AI first, many will die. Many do not understand or consider the economics. Unless you have oodles and oodles of data, you do not make money.
“Are AI agents partners, channels, or parasites? Retailers must choose wisely.”
— Retailgentic, Episode 9
This isn’t a question of if it’s a question of what kind of host you plan to be.
4. The Present: “I can only show you the door...”
“I can only show you the door. You’re the one that has to walk through it.”
We are in the messy middle. Brands and retailers are staring at the door of agentic commerce with mild dread and a lot of PR spin. But someone’s going to walk through. Likely wearing Allbirds. The original ALlbirds, not the new, shittier ones.
4.1 The Interface War Has Begun
This is where it gets bloody. Because the front door of ecommerce is no longer a retailer’s homepage. It’s not even Google.
It’s:
Perplexity
ChatGPT
TikTok’s shopping feed
A browser sidebar powered by Claude or Grok or yet another Elon project
Pick your poison
Wingo’s warning is sharp: if the agent picks where we shop, the brand relationship dies. If ChatGPT finds your yoga mat for you, do you care whether it’s Lululemon or AmazonBasics?And this is changing quickly. It will take a long time to replace Chrome and all its might, but the ascendancy to the throne is happening with more gusto than previous attempts, with more people vying for it and advances on all sides of the throne.
4.2 The Retail Site Is Dead, Long Live the API
Your website is becoming a vestigial limb. If you don’t offer a clean product feed, API integrations, agent-compatible pricing, and real-time inventory hooks, you’re invisible to the AI class.
Retailers now need to act like platforms.
Not in the “we’re building an ecosystem” keynote way. In the “make it easy for non-humans to transact” way.
Because the bots are coming. And they don’t browse. They harvest. Most platforms have started thinking in this way. Headless is no longer a cool term to be used but its principle and reason for being hold true. Shopify want us to think this way. They want their path to be designed for them. No more spoonfeeding from Tobi and Harley.
5. The Organizational Lag: “He’s beginning to believe.”
“He’s beginning to believe.”
You don’t win the agentic race with vision decks and Slack threads. You win it with operational rewiring.
Wingo’s 70/20/10 model first born at ChannelAdvisor is suddenly gospel again:
70%: Keep the lights on (safely).
20%: Test adjacent areas (think plug-and-play agent integrations).
10%: Bet big on wild innovation (aka make friends with your future AI overlords).
Brands that are still 100% in the “we need more banner ads” camp? RIP. This is about dual-tracking your operations: human and post-human. I mentioned ambiguity before. What if timeframes are no longer relevant - think about how different marketing is if everything as a real time chance of being precisely what someone wants at precisely the right time - marketing becomes so efficient it needs only to be fed masses and masses of data. Traffic in the traditional sense is more fractured and split. Was that Vinny shopping or one of his agents. How did our agent react?
Responding is what current agents do, the good ones will react. Use real time intelligence to present better options. But can they do refunds?
6. The Future: “There is no spoon.”
“Do not try and bend the spoon. That’s impossible. Instead, only try to realize the truth... There is no spoon.” ( i have been trying to build this quote in for some 15 years now. This is the second time.)
Commerce isn’t bending. It’s dissolving into a new shape. One where the human is just another variable.
So where are we headed?
6.1 POSITIVE TRAJECTORIES
Agentic Negotiation: Need 10 pallets of pet food at the best rate? Your AI agent is already in a Slack thread with theirs. That agent will also know your bank account, your BNPL status and yoru overall credit history along with your levels of risk. Positive much?
Personalization Reborn: Not just “Hello Vinny,” but “Here’s the vitamin mix you’ll need next week based on your cycling data and fridge stock.”
Commerce on autopilot: From supply chain orchestration to returns processing, the human becomes the editor, not the author. Orchestration, the new word you will hear more than others for some years ahead. Get used to it.
6.2 NEGATIVE TRAJECTORIES
Loss of Agency: You didn’t choose that product. Your bot did. And now your home smells like cedarwood. But it knows cognitive dissonance - in 2 days time you will be thanking your non existent bot.
Platform Power Consolidation: Whoever owns the dominant agents owns the checkout—and everything that comes before it. And JP Morgan are well positioned here. IN previous worlds and thoughts this was to be sacrosanct and the power given to the wetailer was currency. Shopfiy have resisted, maybe they know about this consolidation.
Security Nightmares: Autonomous bots + payment rails = new playground for fraud, scams, and “oops we wiped the warehouse” moments.
Cultural Dislocation: When the commercial experience becomes entirely optimized, it loses its weirdness. Its humanity. Its edge. And somehow we think we will tame it to give us more jobs, better ones. I do not know.
“You have to let it all go, Neo. Fear, doubt, and disbelief. Free your mind.”
We’re being invited to let go of how commerce used to work. Whether we like it or not, someone or something else is now clicking “Buy.”
Final Word: The Choice Is Yours
“This is your last chance. After this, there is no turning back.”
Scot Wingo’s prophecy isn’t that agents might reshape commerce. It’s that they will. The question is whether you build with them, build for them, or get bulldozed by them.
In this new Empire of AI, you don’t sell to a human. You sell to the entity representing them. And if you don’t win that entity’s trust, through feeds, flexibility, and frictionless logic, you’re just noise in a very quiet signal.
So take a moment. Pick your pill. Choose your path.
Because while we’re all busy debating KPIs and click-through rates, the agents are already buying the future.