AI Is Coming... Fast
A clear-eyed look at how quickly artificial intelligence is moving from strange idea to everyday tool.
Quick decoder ring before we start (I throw acronyms around like confetti):
LLM: Large Language Model. The thing behind ChatGPT, Claude, Gemini, etc. You talk to it, it talks back, and increasingly, it does your work.
SWE: Software Engineering. My world.
LOC: Lines of Code. The raw material of software.
A/B test: Running two versions of something to see which performs better.
WFH: Work From Home.
Infra: Infrastructure. The plumbing that makes software actually run.
LLMs have been mainstream for about four years, and they’ve completely redefined entire industries in that window. If I’d slipped into a coma in the summer of 2021 and woke up today, I literally would not know how to do my job. You don’t write code anymore. You tell a robot (which is what I usually call these things) to write the code. Half the time I don’t even look at what it generates until it’s time to review it before it goes live.
Fucking wild, yo.
And this isn’t just a vibe. As of 2026, 84% of developers use AI tools, which now write 41% of all code. Let that sit for a second. Nearly half of everything being shipped is machine-written, and the overwhelming majority of us are riding shotgun with a robot to get it done.
It’s basically a black ball in SWE if you’re not leveraging these tools to the tits. You’re just too slow without them. And time is money.
I Saw This Coming
I’ve always been into tech, and I seem to have this knack for extrapolating the second- and third-order effects of things. I look at what we know today and I feel, in my gut, where it lands tomorrow.
So here’s my read: what I’ve watched happen in SWE is now happening in every industry. Marketing, writing, entertainment, clerical, law, medicine, all of it. And the data backs me up hard. 91% of businesses report using AI in at least one capacity in 2026, up from 78% in 2024 and 55% in 2023. That’s not adoption. That’s a stampede.
Just like you’re now a pleb in SWE if you’re not yeeting thousands of LOC in an afternoon with Claude Code, you’ll be black-balled in marketing if you’re not spinning up 50 social campaigns to A/B test before lunch. The pattern is identical. Only the job title changes.
And the numbers around who wins are brutal in their clarity. Workers who use AI well aren’t just keeping up. Their wages rise twice as fast, their skills evolve 66% faster, and AI-skilled workers command a 56% wage premium. Meanwhile, the World Economic Forum found 40% of employers plan to cut staff in areas where AI can automate tasks. There’s your fork in the road. Pick a side.
The Potato IQ Problem
Here’s what’s been driving me up the wall.
I remember, just a few years ago, basically everyone saying some version of: “Yeah, I tried ChatGPT, it sucks.” Or “It hallucinates all the time.” Or my personal favorite: “It really doesn’t know that much about [my field].”
Now those same people are crying about AI taking their jobs.
How does that make any sense? How do you completely dismiss something as a trivial toy, pointless, beneath you, and then, within 24 months, riot against it? To be honest, I think these people have the IQ of a potato. It was obvious to me what was going to happen back in 2023/2024.
And look, I’ll be fair, because the picture has nuance. There’s even a well-known METR study from 2025 where experienced devs using early AI tools were actually 19% slower, even though they believed they’d been sped up by 20%. The skeptics love that one. But here’s the thing they miss: by early 2026, METR’s follow-up had raw results pointing toward speedup, while explicitly warning that selection effects made the evidence weak. The direction changed in under a year. The tools got better that fast. Betting against that curve is the potato move.
The Lightning Bolt
For me, the moment it clicked was Cursor.
Copilot was cool, but it didn’t show me the future. Then Cursor came out and it hit me like a goddamn lightning bolt. I started using it as much as humanly possible. I made a bet with my time. I bet that this was the future of my industry, that understanding syntax would soon stop mattering, that understanding infra would matter less, and that the only skill that would really count was knowing how to talk to these tools.
Turns out that bet is aging well. In June 2026, SpaceX agreed to acquire Cursor for $60 billion, the largest acquisition of a venture-backed startup in history. The company had crossed a $29 billion valuation and surpassed $3 billion in annual recurring revenue by early 2026. (Congrats to those dudes, by the way. They earned it.) When a rocket company drops $60 billion on a code editor, that’s not hype. That’s the market screaming where the value is.
Now, to be clear, you can’t dismiss everything yet. It’s still genuinely valuable to understand how computer systems and software are architected, and the gotchas that crop up. But that knowledge is being absorbed into the AI itself, fast. Maybe it’s already here, honestly.
The New Flex
So here’s my actual point.
Everyone who works in front of a computer, anyone with a fancy degree and some BS title like VP Senior Executive Director of Analytical Data Foraging, needs to learn these tools. Knowing how to articulate your thoughts and direct AI is becoming the only skill that matters.
Companies are going to shrink. You’re already watching it happen. Research suggests 11.7% of U.S. jobs could already be automated with current AI technology. The people who know how to wield these systems will stay and make absolute fuck-you money, in their careers or on their own ventures. The ones who don’t will simply get chopped up.
This is hitting the white-collar professions everyone assumed were safe, too. Half of all lawyers now call AI a major threat to the traditional practice of law, up from 36% just a year earlier. Seventy-five percent of U.S. health systems use or plan to use an AI platform, and more than half that measured ROI reported at least a 2x return. Law and medicine, the supposedly uncomputerizable fields, falling in line.
It used to be a flex to have 6,000 employees. The new flex is efficiency. The new flex is golf’d. How low can you go and still match or beat your old velocity without the bloat?
The Real Reason I Care
My wife, who I love, has heard me geek out about AI basically every day for three or four years. It’s annoying, I get it. I’ve been trying to get her to lean into these tools forever, and only recently had any luck.
We just had our first kid.
My goal is simple: get us both leveraging AI to do our work faster and better, so we can spend more time with our kid and as a family. We both work from home, and I think this is a giga leverage point for every WFH parent out there. Turn your 6 to 8 hour day into a 1-hour day. Maybe less. That’s not a productivity stat. That’s bedtime stories you don’t miss.
Pick One Thing. Today.
If you’re the “AI person” on your team, you’re going to eat well. You’ll inherit a piece of everyone who got left behind plugging in the telephone routing cables.
So don’t “start using AI.” That’s a New Year’s resolution, and resolutions die in February.
Instead, pick one real task. The thing that actually annoyed you this week. The report you dreaded, the spreadsheet you couldn’t tame, the code that wouldn’t compile, the campaign you didn’t have time to test. Hand that exact thing to Claude, ChatGPT, Cursor, whatever, today. Not to see if it’s perfect. To see how close it gets, and how much of your evening you just bought back.
Do that once, honestly, and you’ll stop reading articles like this one.
You’ll be too busy being the person everyone else is suddenly worried about.
The gap is already open. The only question is which side of it you’re standing on.