AI and who actually wins.
From the failed Tech BisDev Guy
Disclaimer: Here’s the thing. And I’m going to be honest here, I’m not the smartest dude. This is pretty much a raw voice note from a run, cleaned up with the help of the exact thing I’m talking about, so I don’t sound as retarded. My running thoughts are exactly that. Raw. Unfiltered. Unedited. Here they are, polished enough to make some old employer of mine check a box or something for an AI metric so his team can survive until the next layoff round.
I’ve lost track of how many AI narratives are floating around right now. AI is replacing everyone. Build vs buy is back. SaaS is dead. Pick whichever headline you woke up to. None of it is the clean story it’s getting sold as. Build vs buy didn’t come roaring back, SaaS isn’t dead, and AI isn’t strolling in and clearing out the building. It’s all just evolving, and the quick, creative ones are who walk out the far side. And I don’t mean companies and products. I mean you, the person doing the actual work. Imagination is the thing that buys you another day in this.
Developers aren’t getting replaced. The good ones are getting augmented, which is a completely different thing. The “AI is coming for the developers” panic has it exactly backwards. Nobody stopped needing the person. The tool just cranked up how much one person can pull off, and the second it did that, it lit up everyone who’d been quietly getting away with half a job. So let me say the one thing the rest of this hangs on. AI is not killing builders or sellers. It is killing the people who hid behind one half of the work and prayed the other half would never come up. It’s killing the people who basically did nothing but data retrieval, data entry, and fucked around in business applications, reporting, and Excel to justify their existence. Go fuck yourself and your pivot tables.
A model can’t hand you the part that was always scarce. The imagination. A point of view that actually means something. The instinct to walk into a company, find the spot where everything’s grinding, put a finger on the pain quietly bleeding them money, and tell a story about a better version of the place that makes someone with a budget actually care. None of that comes out of a machine. It was the hard part before any of this and it’s still the hard part. Building got cheap, so the thinking and the convincing got worth more, not less. The question is no longer whether you can build it. It’s whether you have any idea what’s worth building, and whether you can get one human being to believe in it.
Who actually wins? The person holding both halves. Not the sales guy who’s all mouth, not the developer who’s all hands, but the one who can sit in a room and hold a vision together, say it like he means it so people move, then go build a rough version of the thing himself. That combination used to be science fiction. The vision people couldn’t build, so they oversold and quietly walked it back when delivery day showed up. The builders could ship anything and still couldn’t make a soul understand why it mattered. That gap is where ideas and budgets went to die, and it’s mostly closed now. Living in that gap is the entire game.
And before anyone runs too far with this, the lazy version of the take is dumb. The pure developer isn’t dead. Neither is the architect. Somebody in a real company still has to make the thing survive contact with production, and that is not a small job. Scale. Security. The integrations into systems built before you were hired. Data that’s filthy. The compliance review that eats a quarter. The one customer environment that detonates everything you shipped. None of that vanishes because some guy vibe-coded a pretty demo over a weekend. He’s still got a scale problem waiting for him Monday morning. That will likely always be there. The deep developer isn’t screwed, not even a little. The SaaS company isn’t screwed. The logic is that the specialists, the support, and the expertise is always needed. Building is great, but thinking you can replace some of these sticky enterprise SaaS products in a long weekend of vibecoding is beyond insanity.
How enterprise organizations function. If you never sat at that table in a meeting, you will never fully grasp the way this works. Organizations are infinitely complex based on structures that refuse to die over decades. The HR departments, the finance departments, a little team you’ve probably never sold to called “procurement.” These honestly are where the greatest product in the world, sold at the most cost-effective pricing models, will still die if the right developers and the right salespeople aren’t navigating that room. From what I’ve seen, no organization has “removed that layer” thanks to AI. So good luck getting your vibe-coded startup into an organization at scale. John from procurement has his own set of compliance, standards, and budget before anything gets signed and implemented. I hope that helps.
Let’s pause and talk about who is screwed.
The one who’s actually cooked is a specific developer. He can build, fine, but ask him why the thing matters, what problem it solves, who’s even supposed to give a shit, and you get a blank stare. That’s fatal, because AI hasn’t laid a finger on the oldest rule there is. People buy from people. They buy from someone they like and respect, someone with a point of view they’ll say out loud and stand behind. They will not buy it from a guy reading features off a screen at them.
Sales splits the same way. The real seller is going nowhere. Trust, taste, a feel for the customer, real context about the industry, the knack for making someone act now instead of next quarter, that is still worth a fortune. It’s the slide-only seller who’s in trouble. Same old snake oil, he just had a nicer deck. When all you ever brought was confidence and a roadmap you couldn’t back up, AI made that act a hell of a lot easier to catch, because the minute anybody can throw a working mock on the screen, charm stops paying the bills. Buyers want to watch the thing run now. A promise doesn’t cut it, and “trust me” was never a strategy to begin with.
Underneath all that sits a whole layer nobody ever fought to keep. The middle. The data-pull guys. Whoever’s job was running the same report every week, copying a number out of one system and into another, chasing down the little stuff no one else wanted to touch. I’ve automated a stupid amount of exactly that work in the last few months. It doesn’t need a human. It needs a prompt and ten minutes to build a skill, agent, or automation with code.
Right next to them, the yes men. Walking, talking versions of “I completely agree with whatever the most senior person just said.” They’ll put on a whole show of having an opinion in a meeting when really they just waited to hear which way the room leaned and then said it back a little louder so somebody noticed them. Zero value added. None. And that nothing they bring is the cheapest thing on the planet now, because a half-decent model agrees with you instantly, and honestly it does the job better, since you can tell it to rip your idea apart and it actually will. That is already more than the yes man ever offered. Show them the door. Repeating the room back to itself stopped being a job.
Here’s the honest part. What you build on your own is a proof of concept and nothing more. It is not a product. It falls over the instant it leaves your laptop. Scalable to a hundred thousand customers? Course not. Going to survive a real customer’s environment, their ancient systems, and their security review? Probably not even close. None of that matters, because it was never the job. The job is that I can show you the thing now instead of describing it and asking you to take my word. The trap is sitting right in front of you. Throw up a slick demo, let the room believe it’s further along than it is, and congratulations, you’re the snake oil guy all over again. Worse, even, because this time the lie actually runs. The move that works is putting the real thing in front of people and being brutally straight about the ninety percent that’s still ugly. Do that, and you become the rare one they believe the next time you walk through the door.
This is where the real corporate rot shows up. Companies are turning AI into a usage contest. Prompt counts. Adoption dashboards. Internal leaderboards. “AI-driven impact” in performance reviews. Everyone wants proof that the workforce is using the magic box. But proof of usage is not proof of value. A guy can light money on fire with API calls, generate seventeen strategy docs, build six demos nobody asked for, and still produce absolutely nothing. No shipped product. No customer saved. No cost removed. No revenue created. Just motion. And motion is the easiest thing in the world to fake.
None of it is a hunch. It’s the documented playbook. Amazon built a system called Clarity to track how much each person uses AI. Meta made “AI-driven impact” a core expectation for everybody. KPMG started grading AI usage in reviews. Amazon set an eighty percent weekly usage target on its own coding tools, and fifty-eight percent of companies now require you to use AI, with ten percent of them firing the people who refuse. Then check what all that motion actually returned. Almost every Fortune 500 tracks usage and almost none track whether a dollar comes back. They’re calling the gap the “AI value illusion.” PwC found ten to twelve percent of companies see any real revenue or cost benefit, and fifty-six percent got nothing. Usage jumped thirteen percent in a year while faith that it even helps fell eighteen.
It’s the same old snake oil, only now it’s the whole org running the con on itself. The slide says we’re an AI company now. The thing underneath it says nothing at all. AI maxxing inside organizations is going to fucking fail. The individual too.
If every developer could turn on local models and have a $1m-a-month side hustle, everyone would be doing it. So no, your AI slop isn’t actually reaching out to every local small business and redoing their website for a $2000 flat rate and making you $50,000 in side money while you sleep. Those charlatans need to die. The age of grift is always the first part of new tech.
I had a business school professor at U of A who hammered one idea into us until it stuck. Goodhart’s law. The second a measure turns into the target, it quits being a good measure. That’s the whole mechanism right there. “AI usage” became the thing you get graded on, so usage turned into the goal, and the only thing that ever mattered, the value, quietly slid off the table.
What actually grinds me down, more than any of it, is that the culture underneath hasn’t budged an inch. Same people, same exhausted playbook, year after year. They hop from one logo to the next hauling the same recycled leadership principles along, reciting them like scripture, and not one real thing changes. Nobody grows. Nobody gets developed. People park themselves in a seat, vest, and cash the check. Some middle managers pocket the credit for work he didn’t do and wraps it up as “my team uses AI more than anyone.” This was supposed to be the jolt that finally broke the inertia, and instead they’re using it to crank out slicker reports about work that still isn’t happening. New tool, same theater. Imagination never shows up on a dashboard anyway, and a place that only ever stares at the dashboard will quietly choke the last of it out of the room.
Pull back and one thread runs through every last one of these people. They fell for a proxy and lost the actual point. Chasing a usage number and never once asking whether anything got better. Polishing a deck instead of shipping a thing that runs. Nodding along when they should have been thinking. Sitting proudly on a leaderboard that points at no outcome anybody asked for. Different costumes every time, same disease underneath. They fell for the fake number and forgot what it was even supposed to stand for.
So, one more time, plain as I can put it. None of this is AI coming for the builders or the sellers. It’s AI coming for everyone who spent a career doing half a job and betting nobody would notice the missing half. That’s the entire story.
Which leaves one thing actually worth doing about any of this. Go find the half of the job you’ve been coasting on and close it. If all you can do is build, learn how to make a room give a shit about what you built. And if all you do is talk, go make something that actually runs so you quit needing somebody to translate for you. The one move that doesn’t survive this is the one most people are still making, parked on their one half, hoping nobody ever bothers to check.
I’m out.
Nick