Happy Monday. Three stories that tie together more than you’d think.
1Naval Ravikant: Software Engineering Isn’t Dead — It’s More Leveraged Than Ever
On Artificial Intelligence (Youtube)
Naval dropped a post over the weekend that became the clearest articulation I’ve seen of what AI actually means for knowledge workers. Not the hype version. Not the doom version. The real version.
His argument: software engineers are now among the most leveraged people on earth. Not because they write code — AI writes code. Because they think in code. They understand what’s happening underneath. And all abstractions are leaky.
I’ve lived this. As a non-technical founder who’s been building with Claude Code and these tools, the bottleneck is exactly what Naval describes: understanding what’s happening underneath the platforms, writing the correct specs, knowing when the AI got something wrong. A strong software engineer isn’t just writing code anymore — they’re writing product specs that AI tools can actually execute on. And they’re catching the bugs, the suboptimal architecture, the leaky abstractions that every AI-built product ships with.
Naval draws a distinction that matters enormously: codified knowledge versus tacit knowledge. Codified is what you learn from textbooks. Tacit is what you learn from experience. AI replicates the first. It cannot replicate the second.
Companies are still hiring software engineers because even if 90% of your code is written by an AI agent, the people prompting those agents and creating those specs have a massive advantage if they’re strong engineers. At the worst, software engineers become the people who fix the problems AI agents create. That’s not replacement. That’s leverage.
The uncomfortable punchline: “There is no demand for average.” But the set of things you can be best at is infinite. “Become the best in the world at what you do. Keep redefining what you do until this is true.”
I see this every day at ModernTax. We build voice agents handling codified, repeatable workflows with government services. But we also have a human component — enrolled agents doing the work. The AI tools help those experienced agents do their jobs better. The more trained and experienced an agent is, the better they perform with the technology. AI amplifies expertise. It doesn’t replace it.
Claude Hit #1 on the App Store — And Nobody Expected What Happened Next
I woke up Monday morning and everybody was talking about Claude. It’s down, it’s back up, but look at these numbers.
The free app rankings tell the whole story: Claude is #1. ChatGPT is #2. Google Gemini is #3. AI has completely taken over the top of the App Store. When you see these three at the top of a global ranking, you understand the sheer scale of what’s happening. This isn’t Silicon Valley in an echo chamber. This is global adoption.
Anthropic went from #42 in the App Store in January to #1 over the weekend. Not purely product — the Super Bowl commercials helped, but the real catalyst was the Pentagon standoff.
Here’s what happened. Anthropic has a $200 million contract with the Department of War. The Pentagon demanded unrestricted use of Claude’s AI models, including for surveillance and autonomous weapons. Dario Amodei, Anthropic’s CEO, said no. The political fallout was immediate: Trump called Anthropic “radical left,” Defense Secretary Hegseth designated them a “supply chain risk to national security,” and OpenAI jumped in to replace Anthropic on the deal.
And then the opposite of what the administration expected happened. #CancelChatGPT went viral. Claude’s downloads surged. The attempt to punish Anthropic turned into the best marketing campaign they never paid for.
There’s a political parallel here that I can’t stop thinking about. Remember 2016? The more liberal and progressive media attacked Trump, called him names, said he was a bad candidate — and it helped his popularity. There’s a correlation with this. The more this administration targets Claude, the more publicity and public support Anthropic gets. The Streisand effect in real time.
My position: I use Claude constantly. It’s one of the best model companies. When you think about Coworking, Claude Code, all these tools — they’re phenomenal and everyone’s using them. Nobody is going to stop. And this weekend proved it.
The Dallas Fed: AI Is Replacing AND Augmenting Workers — But I Think Companies Are Getting It Wrong
The Dallas Fed put out a paper that does something most AI hot takes don’t: it looks at real data. Real wages. Real employment numbers. Not doom and gloom — economics.
The findings confirm what Naval said, but with numbers. US employment is up 2.5% since ChatGPT launched. But in the top 10% of AI-exposed sectors, employment is down 1%. Computer systems design is down 5%.
The decline is hitting young workers hardest — under 25 specifically. Not because of layoffs. Because of a collapsed job-finding rate. Companies just aren’t hiring entry-level in these fields.
But wages in those same sectors are rising. Computer systems design wages are up 16.7% versus 7.5% nationally. Employment down, wages up. How? AI is doing both — automating entry-level tasks and augmenting experienced workers.
The experience premium — the wage gap between entry and experienced workers — correlates positively with AI exposure. High-experience-premium jobs like law, credit analysis, marketing: AI helps, wages rise. Low-experience-premium jobs: AI replaces, wages flat or down.
Here’s where I disagree with the trend. I want to take a short detour here, because ModernTax is hiring new graduates. We’re hiring interns. We’re hiring everybody under 25.
I believe companies have over-indexed on not hiring new grads. Think about it: if you’re building your career right now on the frontier of these AI models, if you’re Gen Z and you understand how young people think, if you’re native to these platforms — you’re incredibly valuable. The right companies understand this. Andreessen Horowitz is hiring people in their teens and twenties because if you’re building a content-focused business, you need people who are on these platforms, creating content, knowing the trends.
The new grad community is an untapped resource right now. Yes, the data says AI is automating codified entry-level tasks. But the people who grow up alongside these tools — who learn to pair AI capabilities with their generational understanding — those people are going to be hugely successful. Companies should be competing for them, not avoiding them.
The Through Line
Three stories. One thread. The value of experience is going up — Naval sees it in code, Anthropic is living it, the Dallas Fed is measuring it. But I’d add a fourth beat: the value of adaptability matters just as much. The new grads who learn these tools aren’t competing against AI. They’re the first generation to grow up as native AI users. That’s an advantage nobody’s pricing in yet.
Tomorrow: SpaceX confidential IPO filing. $1.75 trillion target valuation. The largest IPO in the history of the world, if it happens.
This show is presented by ModernTax — AI and voice intelligence for IRS and tax services. Our agents automate codified workflows, decrease hold times and form processing errors for commercial partners offering small business loans, HR services, and tax and accounting services. 98% accuracy. 75% cost reduction. moderntax.io
Want to go deeper? Reply PREMIUM for the weekly premium newsletter or subscribe below.
