Something Big Is Happening (and I'm in the Middle of It)

Matt Shumer published a piece in Fortune a few weeks ago called “Something big is happening in AI, and most people will be blindsided.” His thesis is that we’re in a February 2020 moment. The signs are everywhere. Most people aren’t paying attention. By the time they do, the ground will have already shifted.

Then this morning, Geoff Huntley’s newsletter lands in my inbox with a subject line that stops me mid-scroll: “Software development now costs less than the wage of a minimum wage worker.” He puts a number on it. $10.42 an hour. That’s what it costs to develop software now. A burger flipper makes more than that.

Shumer is writing from the CEO’s chair at OthersideAI. Geoff is writing from the trenches of the developer community, where he’s been sounding the alarm since he discovered the Ralph Wiggum loop a year ago. They’re both right. And I’m not just reading about it. I’m living it. I’m building AI products with AI, shipping faster than I ever have, and watching the ground shift under the entire industry in real time.

The field report

Here’s what my last month has actually looked like.

I built a remote control for my Mac that lets Claude run commands on my machine from anywhere. I was sitting on my couch, talking to Claude on my phone, and I got frustrated that it couldn’t check a file on my computer across the room. So I built the bridge. MCP, WebSockets, a relay server. I had Claude build it, really. It took an hour. Now I manage repos from my phone while the AI does the typing.

I’m co-building Basic Memory, an AI memory platform that gives LLMs persistent knowledge across conversations. We’re building it with AI, for people who use AI. The tool and the method are the same thing.

At my day job, I built and deployed Fathom, an AI analytics platform for water utilities. It lets operations teams ask questions about their infrastructure in plain English and get answers grounded in live data. I built the platform, the marketing page, and shipped the first version. One person, not a product team.

None of this is hypothetical. It’s just a guy in Austin with a laptop and AI that’s gotten good enough to be a real collaborator.

It used to be impossible to do things like this. Products like Fathom and Basic Memory used to require large teams and serious investment. Now a few people with AI can build at the same scale. The scope of what a small team can accomplish has changed fundamentally.

Where Shumer is right

He makes a point that I think people need to hear: this isn’t like previous waves of automation. There’s no convenient adjacent skill to retrain into, because AI is improving at those skills too. He cites Dario Amodei’s prediction that 50% of entry-level white-collar jobs could be eliminated in one to five years. Shumer thinks that’s conservative.

I don’t know if the timeline is exactly right. But the direction is obvious to anyone paying attention. And his practical advice is solid: pay attention, learn AI, and get your financial house in order. That’s all real.

What he doesn’t say is what it feels like to be on the builder side of this and still feel the ground moving under your feet.

The I-shape

Geoff frames the landscape as a K-shape. I think it’s more of an I-shape. As in: I don’t need an employer anymore. An AI can perform all of the functions my employer had been performing for me. It works both ways. Companies are realizing they need fewer people, and people are realizing they need fewer companies. That’s going to hollow out large businesses from the inside.

The first mechanism is straightforward. Small AI-native teams are going to radically outperform large business structures. Five people shipping what used to require sixty. No middle management. No transformation programs. No committees deciding where AI fits. The one thing that large incumbents always had over small niche players was the ability to execute at scale. That advantage is evaporating. A founder in Geoff’s DMs cut two-thirds of their headcount by simply not backfilling roles after May 2023. Twenty people now produce 30x what sixty did three years ago.

The second mechanism is the one nobody’s talking about. When large companies do layoffs to stay competitive, they push talented people into a job market that doesn’t have room for them. Those people are going to do what skilled people have always done when they can’t find work. They’re going to start their own thing. Except now, starting your own thing doesn’t require a team of twenty and a year of runway. It requires a laptop and an AI subscription. So the layoffs that were supposed to make the big company leaner instead create a wave of small, hungry competitors who know the incumbent’s business inside and out.

That’s not a K-shape. That’s a feedback loop. And it’s going to accelerate.

The honest tension

The execution speed is different now. Radically different. Things that would have taken me weeks are taking days. Things that would have taken days are taking hours. I’m not special. Anyone with a good AI workflow is experiencing the same thing. Geoff describes going to a Cursor meetup where nearly everyone in the room was not a software developer, and they were all showing off things they’d built. They just became software developers because the tools enabled them to be.

I’m bullish on AI. I build AI products. I use AI every day to write code, draft posts, manage projects. I’m betting my career on the idea that this technology makes people more capable, not less relevant.

But I’d be lying if I said I wasn’t feeling the weight of it too. I have friends who are senior engineers struggling to find work. The job market for software developers is the worst I’ve seen in my career. Companies are hiring fewer people and expecting AI to fill the gap. Some of them are right to.

I hold both of those things at the same time. I think AI is going to create enormous opportunity for people who learn to work with it. I also think it’s going to be brutal for people who don’t, or can’t, or who just need more time than the market is willing to give them.

The wizard and the word

Your value used to come primarily from what you knew how to do. Write code. Perform surgery. Draft legal documents. Those skills still matter. But there’s a new layer on top of all of them now, and it’s, “can you drive AI”?

Can you take a vague idea and turn it into a clear prompt that produces something real? Can you direct an agent to build what you’re imagining? Can you spot when the AI is wrong and course-correct before it goes off a cliff? That ability to harness AI as a force multiplier is becoming the skill that sits on top of every other skill. The people who develop it are going to be unreasonably powerful. The people who don’t are going to wonder what happened.

Imagine what it will be like in a few years time where you have a device in your pocket that you can speak with, and it manifests change in front of your eyes. I’m not talking about simple changes like turning on a light switch. I’m talking enormous changes like, “Start a business that does X, and make the website look like Y.” The agents will be able to accomplish every single task from the business filings to the marketing, customer relationship management, billing & accounting, even 100% of fulfillment for digital products.

I think about this in almost spiritual terms sometimes. “In the beginning was the Word, and the Word was with God, and the Word was God.” I’m not trying to be grandiose. But there’s something real happening with language right now. The ability to articulate what you want clearly and precisely has become a kind of superpower. You describe something, and it appears. You write a prompt, and a website materializes. You have a conversation, and working stuff comes out the other end.

Writing well isn’t just communication anymore. It’s creation. The people who can think clearly and express themselves precisely are going to have an outsized advantage in this new world. Not because they’re smarter, but because the machines they’re working with respond to language, and language is a skill you can develop.

If you’re trying to figure out how AI fits into your work, your business, or your career, I’d love to talk about it. And if any of this resonated, reach out. You can find me on LinkedIn, GitHub, or Mastodon.