People Are the Proof

In a pre-pandemic deck, I wrote “best practices are just another way to say boring.” Some higher-up angrily cut the line and said best practices were just that: The. Best.

I look around today and think my line aged pretty well. The everyday web is converging. Everyone is optimizing against the same playbooks. Every social app is the same app. Every checkout is the same checkout. Everything is Amazon, and it’s all a little, well… boring.

This sameness is going even deeper with AI. People use AI to hand them best practices, but it’s just a statistical model optimizing for the most probable output (I asked AI to write that about itself), averaging out everything it’s read. This isn’t AI’s fault: it’s the lack of imagination and emotion on the part of the person prompting it.

The solution? Feeling. I got addicted to Rory Sutherland reels and finally read Alchemy. One of his sharpest points is that differentiation must be an emotion. You have the whole range of human feeling at your disposal, but you must make somebody feel something if you want them to choose you over the competition. Emotion holds people’s attention.

That feeling has to come from somewhere. I called a doctor’s office recently. An AI natural language receptionist walked me through booking an appointment with optimum efficiency. But the experience communicated to me that I’m not worth a human’s time. My call is something to optimize away.

I get it: in healthcare, booking is probably the right part to automate. But what stuck with me was the tone it set for everything after it.

Especially when, a week later, I was in LA and wanted to see the Lakers. Couldn’t get tickets so went to Cosm to watch the game on a screen. A screen so huge and immersive, that I knew instantly I’d never go see basketball in person again. 

LeBron was sweating and joking a few feet away, larger than life and in high definition. The screen took a human being and amplified him. Every feeling in the room got bigger because the humanity was made bigger.

Same underlying technology. Opposite effect. The difference was whether the experience became more or less human.

For 15 years, we’ve removed distractions, cleared obstacles, sped customers to conversion. But our success? It’s less for the user, and more for the “job to be done.”

But friction is life. Friction is literally how we hold things, how we hold people. Pleasantries, surprises, figuring out where the milk is in the store, all friction, all optimized out of online transactions because hey, I just need filters for my cat’s water fountain and I need them tomorrow.

And for the cat fountain filters, that’s the right call. Nobody wants a relationship with a checkout flow when the checkout is a minor touchpoint.

But what happens with agentic commerce when the AI owns discovery, comparison and recommendation? When your relationship with a customer may only be a checkout flow? 

When even life decisions are mediated through AI, where are you still human? Where do you make somebody feel something? Where do you amplify a person instead of optimizing them away?

Today, there’s pressure from every side, from every dashboard. Remove a step and watch conversion climb in tandem. Add the right kind of friction and the payoff is a renewal two quarters out, in a metric nobody’s bonus is tied to. We built companies that can measure the cost of trust but not the value of it.

There’s a balance. A good mechanic wants to turn a profit and build trust, so he shows you the worn part after he replaces it. When people’s livelihoods depend on transactions, they manage to make the transaction feel human. We used to do that online. We optimized it away.

Show the people behind the AI: a name attached to the work, a real person’s review, a founder who signs what they ship, a “talk to a human” five minutes in. Visible evidence that someone with a reputation stands behind the experience and the customer’s time is valued.

Then, like the mechanic, show a bit of the work. Provide sensory feedback. Build in the right kind of friction. Especially if the AI works out of sight. 

AI is very good at sounding like us now. The receptionist who booked my appointment had a warm voice, natural pauses. Most people can’t spot the difference. In one widely cited study, a chatbot’s answers to patient questions were rated more empathetic than real doctors’ answers. Then researchers ran the flip side: told people a machine wrote the message and the exact same words ran cold.

So you can fake emotion, for a while. But you can’t survive the reveal. Every synthetic “sincere” apology is one screenshot away from becoming the story about your company, and the market is now full of people and agents whose favorite hobby is the reveal. Sounding human and being trusted are different things.

I’m talking about a different kind of trust. Software can be reliable (we trust calculators all day). But the trust that builds a business is accountability: a person’s name on the line if the thing fails. The mechanic shows you the worn part because his reputation sits with it on the workbench. A signature means something because it costs the signer. Real life has a handshake; AI has to offer equivalent proof. That’s the one thing it still can’t fake: not a warmer voice, but something human at stake.

1. Where in this does a person actually show up?
2. When it fails, is there something human in the response?
3. Can the user watch it work, or can they only wait?
4. What here could a competitor rebuild by Friday, and what could they never copy at all?

That last question is the most important. Your idea is already out there. Your interface is easily copyable. But the part that’s uniquely yours are the people behind it, the human story that led to its creation, and (dare I say) the best practices you willingly ignore.

Everyone can build the same product. Prototypes are a commodity. One thing that doesn’t converge, however, is the credibility of the people behind the work. Build the intelligence, then show us the humanity. That’s the part no one can fake.