I saw an ad on LinkedIn the other day for a startup that's going to revolutionize golf with AI.

The pitch was thorough. The app would book your tee times based on your known preferences and the people you usually play with. It would optimize for the conditions you score best in — wind direction, green speed, time of day, course layout. It would scan video of your swing and give you tips. It would coach you through your round in real time. It would manage everything around the act of golf — except, mercifully, the act of golf itself, which it had not yet figured out how to swing for you.

I sat with that ad for a minute. I'm a returning golfer, four years back into the sport after a sixteen-year hiatus, currently working with a human coach to get my swing back. I am, in theory, the target customer for this kind of product. I am the guy who would benefit from a tool that gets me onto the course faster, in better conditions, with the right people, using the right club.

I closed the ad without clicking it.

Not because the technology was wrong. The technology will be fine. The technology will, in fact, be excellent. I closed it because the pitch made a quiet assumption about why I play golf, and the assumption was wrong, and the assumption is the same one a lot of products like this are about to make about a lot of things — and I think it's worth pulling apart before too many of us walk into it without noticing.

The frictionless promise

The pitch every productivity tool makes, in some form or another, is we will remove the friction. This is true of calendar apps, food delivery services, AI writing tools, fitness trackers, learning platforms, and now, apparently, AI golf coaches. The friction-removal industry is enormous, growing, and largely unexamined. We have so completely accepted the premise that less friction is better that the business model writes itself. Whatever the activity is, the pitch is identical: we will give you the result with less of the in-between.

There is a category of friction for which this is genuinely good news. Booking a flight used to take a phone call and a travel agent. Filing taxes used to take an accountant and a stack of receipts. Writing a brief used to take three associates and a week. AI is going to remove a great deal of friction that was just bullshit between us and the thing we were trying to do, and we should welcome that. I run an MSP. Removing friction between professionals and their actual work is, in some sense, the entire business I'm in.

But there is another category of friction, and we are about to get this category badly wrong if we treat it the same way. The other category is the friction that is the thing.

Golf is in the second category.

The piano in the corner

The cleanest way to see this is to think about a player piano.

We figured out how to make pianos play themselves more than a century ago. The technology was impressive. You could buy a roll, feed it into the mechanism, and the keys would depress in perfect sequence to produce, mechanically, the same notes a pianist would have produced. Plenty of homes had one. They were, in their way, the AI of their era — a music-rendering machine that removed the friction of having to know how to play.

Nobody, in the entire history of the player piano, ever confused the player piano for a pianist.

Nobody said well, the music is the music — what does it matter who's making it? Nobody asked whether they should still bother learning the piano now that the player piano existed. Nobody argued that the player piano was just an optimization of the pianist. We understood, intuitively and without having to debate it, that the playing was the thing. The music was downstream of the playing. The playing was where the meaning lived. A player piano produced the sound, but it didn't produce the experience of having played.

A century later, we somehow have to relearn this lesson, in real time, against a pitch that has gotten dramatically better at obscuring the question.

The AI golf app produces the score. It does not produce the round.

The line between optimization and replacement

I want to head off a fair rebuttal before it lands, because I'd make it myself if I were reading this.

I have a human golf coach. I'm taking lessons. That, too, is a form of optimization — paying somebody else to teach me what I would otherwise have to figure out on my own through trial and error. If the argument is that golf is supposed to be hard, isn't a coach also cheating? Where's the line?

The line is between help that gets me to the doing and help that replaces the doing.

A coach is in the first category. He stands on the range with me, watches me hit a shot, tells me my hips are spinning open before my hands square the club, and gives me a drill to fix it. The next time I stand over the ball, I'm still the one swinging. I'm still the one who has to feel the difference between the old motion and the new one. I'm still the one who has to repeat it a thousand times until it lives in my body. The coach made the path shorter. He didn't walk it for me.

The AI swing-bot — the one that simulates the perfect ball flight and tells me my round would have been a 78 if I'd hit each shot the way the model recommended — is in the second category. It produces the output of having played a great round. It does not produce the experience. The score is real. The playing is not.

It turns out almost everything in modern life sits somewhere on this spectrum, and the question worth asking is not is this optimization good but is this optimization in service of the doing, or is it replacing the doing? Booking a tee time through an app that knows my preferences is in service of the doing. Reviewing the conditions data so I show up to a course where I'm likely to play well is in service of the doing. An AI that "plays" my round for me by rendering what would have happened if I had hit better shots is replacing the doing, and it doesn't matter how good the rendering is, because the rendering was never what I came for.

Some friction is just bullshit between you and the thing you're trying to do. Some friction is the thing. Knowing which is which turns out to be the entire game.

What I actually want from a round of golf

I am 44. I returned to golf four years ago after a sixteen-year hiatus, and this time I came back to actually learn the sport — not to ride around in a cart with a cooler of beer and call it a hobby.

What I want from a round of golf is one shot.

Not eighteen good shots. Not a great score. Not a personal best. One shot, somewhere in the round, where the swing was the swing I was trying to make and the ball did what I asked it to. Or sometimes, even better, one shot where I had no business pulling it off and it worked anyway — the chip off the side of the hill that rolls into the cup by accident, the recovery from a bad lie that nobody on the course saw coming, including me. The drive home is built on that shot. That one is why I came back next week. The other 99 swings can be whatever they were. I don't need a perfect round. I need one moment of I meant that.

Now look at what the AI golf app is offering me.

The pitch is that it'll review every shot — not just the great one — and tell me what I did wrong on each. My hips. My follow-through. My shoulder turn. My weight transfer. My grip pressure on the seventh hole. The release point on the fourteenth. The path of the club at impact on every shot the watch tracked. By the end of the round, the app has 99 things it wants me to work on. None of them is the chip off the hillside. The chip off the hillside doesn't show up in the model, because it was a recovery shot the system didn't expect and can't credit.

Read what just happened there. The product whose entire pitch was removing friction from my golf game has, by being thorough, generated 99 new sources of friction. Every shot I didn't think about now has an asterisk and a coaching note. Every imperfect motion is now a problem with a recommended drill. The drive home — the one that used to be carried by the chip off the hillside — is now carried by a list of 99 things I need to fix before next Saturday.

That isn't optimization. That's a different sport, with a different reward structure, that the app has quietly substituted for the one I came for.

The score is real. The round was not.

What we offload at Indevtech, and what we don't

I'd be a hypocrite if I didn't acknowledge that this same distinction operates in my actual job, and we make these calls every day.

What we offload at Indevtech is the bullshit that was never the right use of a human mind to begin with. Routine documentation. Alert triage. The kind of formatting and reformatting that used to eat hours of an engineer's week. Status reports. The administrative undergrowth of a managed services business is enormous, and AI is genuinely freeing my staff from work that nobody, in twenty-five years of running this company, has ever told me was the meaningful part of their job.

What we deliberately don't offload is the part where we look at a client and think about what they actually need. The part where someone exercises judgment under uncertainty. The part where we write something in our own voice — an email, a strategy memo, a recommendation — because the voice is part of why the client trusts us. I have a personal rule, mostly self-imposed, that I will not draft certain kinds of writing through AI even if AI could plausibly produce something close to what I'd write. The risk is too high that the AI's version becomes my version through suggestion, and the next thing I know, my voice has been quietly assimilated into the average voice the model was trained on. Some things are worth the friction of writing them yourself, even if the friction is annoying. Especially if the friction is annoying. The annoyance is part of how you stay you.

The freed-up brain isn't a bonus. It's the whole point. Because of the work AI is now doing for us, I get to spend more of my time with my clients looking them in the eye. I'm not flipping between six tabs trying to assemble notes for a call. I'm sitting across from somebody hearing what they're actually telling me. The friction we removed was bullshit friction. The friction that's left is the friction that is the work — the relationships, the judgment calls, the moments where being a present, thinking human is the value. We removed friction so we could spend more effort in the places where effort matters.

Which is the same move I'm trying to make on the golf course.

The phrase that needs an autopsy

There's a phrase most people use without examining: work-life balance.

The way it usually gets deployed implies a kind of seesaw — work on one side, life on the other, and the goal is to get more of the life side and less of the work side. Less time at the office. Less stress on Sundays. Less email after six. Less effort, broadly construed.

I don't think this is the right model.

I think the people most strung-out on the work-life-balance question are not, in fact, working too hard. I think they are people who have made everything outside of work so frictionless that work is the only place left where they exert themselves at anything. They scroll their hobbies. They DoorDash their dinners. They half-watch their movies. They consume their relationships through asynchronous text. They have so thoroughly removed effort from the rest of their lives that work, by elimination, becomes the entire load-bearing column of their identity. And then they wonder why they feel imbalanced.

The fix is not less work.

The fix is more chosen effort outside of work.

I exert effort at my actual job, and I am happy to. I also exert effort, deliberately, at things that are not my job — golf, bonsai, writing this article, returning to a sport I was bad at and getting incrementally less bad at it, in public, in front of a coach who watches me chunk a wedge into the side of a hill and then patiently explains the next drill. I do these things, in part, because I want to remind myself that effort is not something I reserve for the office. Effort is a chosen good. A prize, even. The thing you do because the doing makes you feel like a person.

If you have arranged your life such that work is the only place you exert yourself, and you are wondering why your life feels off, I would gently suggest that the problem isn't work. The problem is that the rest of your life has been quietly defriction-ed into a kind of well-upholstered nothing, and you are starving for the experience of effort somewhere else.

The frictionless pitch — the one that says we'll remove the friction from everything in your life so you can rest — has been, for a lot of people, slowly producing the opposite of rest. It's been producing a population that has no idea what to do with itself when nothing is being demanded.

Golf is hard on purpose. Bonsai is slow on purpose. Writing in your own voice is awkward on purpose. The hard, slow, awkward parts are not the cost of admission. They are the admission. The whole point of doing those things is the doing.

I'm 44. I'm playing golf again, badly, with a human coach, in conditions I don't get to optimize, on a course where the wind sometimes ruins my round, against partners who occasionally outdrive me by 30 yards. I am, by every measure the AI golf app would track, sub-optimal.

I am also, by the only measure that matters to me, exactly where I want to be.

The doing is the point.