Time to catch up.
We’re in the “made by humans / don’t use AI / I can tell you used AI” phase. Some companies are waiting for a “killer business feature.” Meanwhile, OpenAI’s latest releases point to a route that always pays: fantasy and pleasure. If enterprise stalls, entertainment rarely does.
A few weeks ago, consulting on a product story, my heart sank. The ask was familiar: turn a vague feature description into something tangible, marketable, useful. I’ve done this my whole career.
I ran my raw prototyping loop: add context from years in product plus three years of product-oriented prompting, run it through a PM persona, sanity-check in another tool, then hand it to an AI coding tool for a first pass. Only then do I design — prototyping in tight loops instead of rearranging atoms from a design system. It’s faster to learn by making.
That’s when the existential crisis struck: a vertical slice I’d spent years learning could be done by a few apps in minutes. Watching competence turn into a button is disheartening. Then I remembered: product design is a slice, not the cake. I’ve had a few careers. Reinvention isn’t a bug in my life, it’s a habit.
At Field I told the team to use AI early and often. Not because everyone has to, but because literacy changes what you see. Once you use it, you spot it in other people’s work. You learn which app fits which contour and how prompting shifts by model. The leverage moved: it’s no longer in using AI, it’s in using it with intent and precision.
There’s also a perception trap. People say the models are getting dumber. Mostly, our expectations got sharper. Proficiency raises your floor and your standard. What felt magical last quarter feels basic after 200 hours in the tool.
Bottom line: AI is a near-perfect instrument for late-stage capitalism: near-infinite content at near-zero marginal cost. That doesn’t make it evil (yet), it makes it predictable. Systems do what they’re wired to do. Plug a content synthesizer into an attention market and you get volume. Plug it into a team with tight feedback loops and you get learning. Plug it into lonely hours and you get fantasy. Same electricity, different circuits.
Know which game you’re playing and pick the move that fits the game. If the game is efficiency, play it openly. Use the tool. Measure output and error rates. Ship. If the game is authenticity, own the friction. Slow down. Write the paragraph yourself. Say what only you can say. If the game is expression, mix the two. Let the model be your bandmate.
Not every context deserves your full, handcrafted self. That’s how burnout dresses up as virtue. And the button isn’t always better. There are stakes — trust, taste, and the long memory of the people who rely on your work. Before any project, I ask: what’s scarce here? If time is scarce, I automate and compress. If trust is scarce, I slow down and sign my name. If imagination is scarce, I prototype until we see the edges.
Tools change the slope of the hill. They don’t choose which one you climb.