the headset is not the moat

when i taped with this founder, he handed me the episode title before i could think of one. the headset is not the moat.

he builds a vr and ai platform for clinicians who work with kids with autism and other neurodevelopmental conditions. it's in more than a hundred and fifty clinics, with over twenty three hundred people reached, by his numbers.

and everyone who sees it stares at the headset. it's the shiny part. it's the part you'd put in a pitch photo.

he thinks that's the least defensible piece of the whole company. anyone can buy a headset.

if you've ever asked what a data moat is for a startup, his answer is the clearest one i've heard this year.

start with the clinical problem. it's called the generalization gap. a child learns a skill in the clinic, crossing a street, ordering at a counter, and then the skill doesn't transfer to real life. the clinic is quiet and controlled. the world isn't.

vr helps because it can strip out distractions at first and then simulate real places, step by step. sessions run fifteen to twenty five minutes. he cited therapy timelines shrinking in some cases, one example going from seven months to two.

but the headset only delivers the session. the moat is what gets captured around it.

there are three streams. first, what happens inside the vr session. second, the clinician's notes. third, a parent app that captures what happens at home... did the skill show up in real life or not.

as i put it back to him on air, "so you know you said that the headset is not the moat. right, that the behavioral kind of data loop is."

the ai sits on top of all three and learns which activities produce results outside the clinic, not just inside it. "so generalization gets faster. the development gets faster."

that's a data moat. not a big pile of data. a loop. the product creates data, the data makes the product better, the better product brings more clinics, the clinics create more data. every turn of that wheel makes the next competitor further behind, because they'd have to rebuild years of real world outcomes, not a piece of hardware.

it's like a family recipe. anybody can buy the same pot. the reason grandma's soup is better is fifty years of adjusting the salt based on who came back for seconds.

a lot of founders tell investors their moat is their technology. in an ai world that's getting weaker every month. models get cheaper, hardware gets cheaper, and somebody copies your demo by the weekend.

what doesn't get copied is outcome data that only exists because you were in the room, across hundreds of rooms, for years. especially when one stream of it comes from the real world, like his parent app, where no simulation can fake it.

so if you're building, ask the harder question. where's the loop? what does every customer give you, just by using the product, that makes it better for the next one? and could a well funded competitor buy their way to it in a year?

if they could, it's not a moat. it's a head start.

and notice the order he built it in. the session came first, because nobody pays for a loop. clinics pay for a tool that helps a child today. the loop is what the tool leaves behind.

i wrote more on building ai companies around data nobody else has in the ai method.

so what's your headset... the thing everyone looks at, and what's the loop behind it that nobody can buy?