Sell before I build

I spent most of today talking to clients, which is still a slightly strange thing for me to say because I’m usually much more comfortable disappearing for a few days, building something, and then coming back with a product I can point at. Today was the opposite. I spent the morning talking to one person, the afternoon talking to another, and then somehow the rest of the afternoon talking to more people. Most of them still don’t really know exactly what my product is, and I’m realizing that, at this stage, that might actually be fine. I know the general problem I care about, I know what I think is changing, and I have a fairly strong thesis about where things are going. What I don’t know yet is the exact shape of the thing people will pay me for, and I’m trying not to pretend that I do.

This is probably the biggest thing I’m doing differently this time. I’ve spent a lot of time in the past confusing people finding an idea interesting with people actually having a painful enough problem to buy something. It is incredibly easy to get someone to say, “That’s so cool,” especially when you’re talking about AI. It is much harder to get to the point where they start describing the annoying, specific thing they deal with every week, what happens when it breaks, who gets blamed for it, how much it costs them, and what they already tried to do about it. Those conversations are much less glamorous than building, but I’m starting to think they are the actual work.

So my motto this time has become sell before I build, although even “sell” is probably too strong a word for what I’m doing right now. I don’t want to convince someone to buy a product I invented in my head and then force their problem to fit inside it. I want to understand the problem well enough that I can eventually promise something very concrete and know I can deliver it. My current MVP is almost comically unimpressive: it is basically a website explaining the thesis and asking people to get in contact. There isn’t some giant platform hiding behind it. There isn’t a beautiful dashboard with twenty features that I spent three months building. There is an idea, a waitlist, and me talking to people. I did, obviously, make the website extremely agent-friendly because it would be ridiculous not to given what I’m working on, but otherwise it is probably the least product I have ever had while trying to start a company.

Later today I went to buy a book as a gift and ended up choosing *Apple in China*, and it felt weirdly connected to everything I had been thinking about. A company like Apple can have one of the strongest brands, products, and operating systems in the world and still find that entering a different environment changes the rules underneath it. China has its own incentives, regulations, supply chains, political relationships, consumer behavior, and dependencies, and Apple had to become deeply entangled with that environment in ways that would have been hard to predict from Cupertino. Something can work incredibly well in one place and still require a completely different way of operating somewhere else.

I think startups are like that on a much smaller scale. You can read a hundred stories about how another company found product-market fit, copy the sales motion of a startup you admire, use the same pricing model, or build something because it worked for a different type of customer. You can even be technically correct about the future and still be wrong about what somebody needs from you today. The environment matters. The customer matters. Their constraints matter. And those things are difficult to learn from your laptop.

So for now I’m trying to resist my strongest instinct, which is to build the answer before I have finished asking the question. I have a website, a thesis, a lot of conversations, and not much else. Somehow that feels more legitimate to me than having a beautiful product nobody asked for.