Pith — Why We Build

Most people I talk to have already made up their mind about AI, and it splits pretty cleanly into two camps. One thinks this is the thing that finally revolutionizes how we work and live. The other thinks this is the thing that finally comes for their job. Both camps are loud. Both are pretty sure they're right. The answer is probably both. And also neither.

Because the reality is, AI is just a hammer, and what it enables us to do depends on who wields it.

Sure, every hammer is slightly different — length, weight, features — but at the end of the day, they all hit a nail.

And so the challenge is, when you have the world's greatest hammer, you spend all your time looking for nails. Even things that look like they maybe, kinda, possibly could be nails if you squint your eyes just right. Because you have this incredible hammer and, damnit, you want to use it. And if you don't use your hammer, then someone else is going to come along and use theirs, so you try and use it as much as you possibly can, even when, deep down, you know it's the wrong tool for the job.

And just like a hammer, AI is something anyone can pick up now. The differentiator was never having one. It's what you build with it.

I tested this on myself before I ever tested it on a product. I was a product leader looking for more out of these tools — same access everyone else had, nothing special about my situation. What actually moved the needle wasn't the AI getting smarter. It was what I started bringing to it: the structure, the context, the rigor of what I gave it before I ever hit enter. I pointed that at the tedious, high-friction parts of my own job — the stuff eating time that should have gone to actual thinking. Output went up. Time went down. None of it came from a better tool. It came from me getting more deliberate about what I gave it.

That's the part almost everyone skips while hunting for nails. The hammer was never the edge: what you build with it is.

There's something worth naming about making this argument in an essay developed with AI assistance. I'm not going to pretend otherwise. That's part of the point, and I'll come back to it.

There's a tension every product leader lives with, between thinking at the right strategic altitude and staying close enough to the work to maintain quality. Agentic tools started collapsing that gap for me faster than I expected — not by replacing judgment, but by translating strategic intent into execution fast enough that I stopped having to choose one or the other.

Here's the part I said you build around it. Same rigor, aimed at someone else now — baked into the product instead of something they have to learn themselves.

The same technology a Fortune 500 strategy team is running is available to a solo founder with a subscription. Access stopped being the differentiator a while ago.

What isn't commoditized is the structure around the capability. What questions to ask. What context actually matters versus what's noise. What guardrails prevent the output from being confidently, plausibly wrong in ways that are hard to catch. What design makes the result usable by the person who needs it rather than just the person who built it — which are not always the same person, and the gap between them is where a lot of otherwise decent AI products go to die.

Some of my read on this comes from watching my mother work — she was an editor and designer by trade. Most of it comes from years spent moving across industries and organizations that were solving the same problems in different ways, which builds something I'd call a negative knowledge base: you've seen enough implementations fail that you recognize the failure mode before it finishes happening. It isn't taste. It's pattern recognition from exposure, and I trust it more than most people trust their own instincts, because I've tested it against enough variations to know when it's right.

The difference between an AI-powered product that changes how someone thinks about what AI can be for them, and one that confirms every skepticism they already had, is usually not the AI. It's everything around it.

I want to be honest about something: the cynicism a lot of people feel about AI right now is earned, not manufactured. I know because I've had it myself.

Some of that cynicism isn't even about the technology. It's about how quickly some companies started treating anyone who hadn't caught up yet like that was a character flaw, instead of a completely normal place to be with something this new.

Because... well, shit, no one's an expert at first!

A few years ago I was in a leadership position where I could see this technology starting to change how work gets done — starting to revolutionize whole categories of business — and I didn't have the organizational leverage to do anything with that. First-world problem, and I know it. That's not the same as actually losing a job to automation, a real micro-economic loss even when the macro-economic case for the change is completely sound, and I don't think we do people any favors pretending that tension isn't there. But what I had was a real, personal version of watching something enormous happen without me, wondering if I'd already missed the window. That's what sent me looking, on my own, at what these tools could actually do.

So when I say the coverage skews toward displacement and risk, and that if that's your primary exposure to AI, cynicism makes sense — I mean that as someone who's been there, not someone diagnosing it from a distance.

Here's what I'd add, though. Cynicism you've earned is still cynicism you're paying for. It keeps you from the firsthand experience that would actually update your view. What finally moved me wasn't an argument. It was the experience of the tools actually working, well enough that I couldn't keep telling myself the old story.

I want to be precise here, because it's easy to oversell this. Satsuma isn't trying to fix macro displacement — that's a policy problem, beyond what a venture studio does. We're not running digital literacy programs. What we're doing is narrower: build products well enough, and design them honestly enough, that the people who use them come away with a different sense of what's possible for them. Not because we told them. Because they used the thing.

That's the whole bet, actually. Not "understand AI better." Not "be less afraid of it." Just: feel like it's yours to use, instead of something being used on you or around you. Powerless to empowered, one product and one experience at a time. It doesn't sound like much next to the size of the problem. I don't think that makes it small.

What I found, when I finally had the space to go deep on this, is that AI had more power than I understood — and that most people hadn't had the chance to find that out themselves, not because it was inaccessible, but because nobody had built the right structure around it.

Someone will eventually. The moment is real. Why not now. Why not us.

That's why we build.