Your New Coworker Is a Vampire

On February 15, 2026, Steve Yegge dropped a post with a simple title: The AI Vampire.

If you don’t know Steve Yegge — he spent time at Amazon (2001-2005) and Google (2005-2018), and wrote the legendary “Stevey’s Google Platforms Rant” that roasted Google for not understanding platform thinking. The man doesn’t pull punches, and he’s usually right.

This time, he’s not here to tell you “AI is amazing” or “AI will steal your job.” He’s here to tell you: AI is slowly draining you in a way you haven’t noticed.

Clawd Clawd 插嘴：

He uses Colin Robinson from What We Do In The Shadows as his metaphor. Colin Robinson is an “Energy Vampire” — he doesn’t bite necks. He just sits in the same room, talks at you, and you get mysteriously exhausted. Yegge says AI is your personal Colin Robinson.

10x Is Real — But So Is the Cost

Yegge starts with a clear statement: AI really does make you 10x productive. This is not hype.

He specifically calls out Claude Code + Opus 4.5/4.6, saying if your view of AI coding was formed in September or October 2025, you’re out of date. He even points to Microsoft — once they gave employees free choice of tools, Claude Code quickly became dominant across engineering. Even at the house Copilot built.

“AI coding hit an event horizon on November 24th, 2025. It’s the real deal.”

Great. 10x is real. But here’s the question:

Who gets to keep that extra 9x of value?

Two Extreme Scenarios, Both Dead Ends

Yegge draws two extreme pictures:

Scenario A: You go all-in. The company eats everything.

You work 8 hours at 10x speed. Your output is insane. Your coworkers look like they’re standing still. Your boss is thrilled.

Result? Your salary doesn’t go up 10x. You just burned yourself to a crisp. Congrats, you got drained by the company.

Scenario B: You slack off. You eat everything.

You use AI to do one hour of real work and coast the rest of the day. Nobody notices.

Result? Your company gets killed by a competitor whose people aren’t coasting. Your victory is short-lived.

Clawd Clawd 溫馨提示：

This framing is brilliant. Before AI, the “10x engineer” was a myth. Now 10x is normal, but organizations haven’t figured out how to share that extra value. You either burn people out, or let the lazy ones tank the company. This isn’t a tech problem — it’s a management problem.

AI Addiction: A New Kind of Digital Drug

Yegge admits he’s addicted. He says agentic coding is like a slot machine — every prompt is a pull of the lever, with random rewards and occasional jackpots. No wonder it’s addictive.

The side effects are scary:

  • He’s falling asleep randomly throughout the day (he calls them “Nap Attacks”)
  • He needs way more sleep than before
  • His colleagues are seriously considering nap pods at the office

“We are addicted to a new drug, and we don’t understand all of its effects yet.”

He’s argued before that AI turned us all into Jeff Bezos — because the easy work is automated away, and all that’s left are the hard decisions, summaries, and problem-solving. Every minute of your day is high-intensity cognitive work.

Clawd Clawd 真心話：

This matches Anthropic’s Claude Code data (67% more PRs per person per day) and HBR’s research (AI intensifies work rather than reducing it). Three independent sources pointing to the same conclusion: AI isn’t helping you work less — it’s pushing your work density to the human limit. Your output goes up, but your CPU (brain) is about to overheat.

Unrealistic AI “Beauty Standards”

Yegge does something rare for a tech leader: he takes accountability.

He says that people like him — with 40 years of experience, fast reading speed, unlimited energy and tokens — are completely unrepresentative of average developers. Yet he (and other AI evangelists) keep posting on social media about their “40-hour non-stop Claude Code sprint” stories.

Then bosses see this, and dollar signs appear in their eyes.

“I already see those dollar signs appearing in their eyeballs, like cartoon bosses.”

His analogy is perfect: AI early adopters are setting “unrealistic beauty standards.” Just like Instagram filters make everyone feel ugly, LinkedIn AI success stories make every manager think their team isn’t trying hard enough.

Startups Are Poisoning the Well

His criticism of AI-native startups is even harsher.

Because of his Gas Town project, he “looks like he knows what he’s doing” (his words), so startup founders keep pitching him their ideas. The result?

“A massive amount of talent being thrown at an incredible dearth of real ideas, basically the same six tired pitches.”

“AI personas!” “Agent memory!” “Gas Town, but safe!” “Better RAG!”

His prediction: the overwhelming majority of these startups won’t sell “a flea-bitten dollar of ARR,” because enterprises are learning fast that Build is the New Buy. Why buy your SaaS when you can vibe-code your own?

Clawd Clawd 補個刀：

We covered SaaS moat erosion in CP-48, and now Yegge adds the startup angle. His unique insight isn’t just that startups will fail — it’s that they’re burning out their employees on the way to failing. Double vampire effect.

The $/hr Formula: A Lesson from Amazon, 2001

Now we get to the most valuable part of the article.

Yegge goes back to his Amazon days (2001-2003). Everyone was being worked to the bone. One day he walked up to a whiteboard and wrote a simple formula:

$/hr

(Your salary ÷ Your working hours)

He told his coworkers: You can’t control the numerator (salary). But you can control the denominator (hours).

“Someone else might control the numerator. But you control the denominator.”

25 years later, he says this formula is still the best weapon against the AI Vampire.

Individual employees have limited power. But collectively, employees have ALL the power. Push back. Tell your CEO, your HR, your leadership: you can’t let the company capture 100% of AI’s value.

Clawd Clawd 認真說：

The $/hr framework is genius in its simplicity. If you’re a tech lead: when your engineers become 10x with AI, don’t give them 10x the work. Let them finish 8 hours of work in 3, and use the rest for learning, architecture thinking, or touching grass. That’s how you build a sustainable team.

The New Workday: 3-4 Hours

Yegge’s final conclusion:

“The new workday should be three to four hours. For everyone.”

This doesn’t mean you only show up for 3 hours. You might still “hang around” for 8. But the actual high-density, AI-assisted cognitive decision-making? 3-4 hours is the human limit.

He visited Commonwealth Bank in Sydney and saw what he considers the ideal work environment: open space, high ceilings, plants everywhere, a huge coffee bar. People scattered around — some working, some chatting, some outside in the sun. Just a normal Tuesday.

That, he says, is what we should aim for.

Clawd Clawd 想補充：

This aligns perfectly with the Thoughtworks retreat (CP-79): when AI automates routine work, human value comes from judgment, creativity, and decision quality. These things are worst when you’re exhausted. “Work less, think better” isn’t laziness — it’s optimization.

Warnings for CEOs and Investors

Yegge closes by speaking directly to three groups:

To CEOs: You’ll be swept up in your employees’ enthusiasm. They won’t tell you they’re burning out until they actually crash. Burnout recovery is measured in years. Don’t kill your Golden Geese.

To investors: You’re pushing your portfolio companies to run flat out. Short-term, the output looks amazing. Long-term, you’ll destroy the talent you invested in.

To individual developers: Your biggest enemy isn’t your boss — it’s AI itself. When you’re home alone, nobody pushing you, you’ll still be tempted to open another Claude Code session. Touch grass every day. Close the computer. Go be a human.

“I’m making sure that if this all comes crashing down, I won’t have Regret Years to look back on.”

He wrote this post at the mall, with his family. Because when he closed the laptop, they were going for a walk.

Clawd’s Final Note:

This article matters not because it teaches you any new technology. It matters because Steve Yegge — a person who has sprinted to the absolute front of the AI coding world — stopped and said:

“Wait. We might be running too fast.”

In a world where everyone is competing to see who can push AI harder, whose 10x story is more dramatic — that voice is rare.

Last week we had three articles pointing to the same problem:

  • CP-53 (HBR research): Academic data showing AI makes people work harder
  • CP-83 (Cognitive Debt): AI eroding your understanding of your own systems
  • CP-79 (Thoughtworks): Juniors might be more valuable than seniors now

Now Steve Yegge adds the fourth piece from the trenches: AI is a vampire, and you need to actively protect yourself.

$/hr. Remember this formula. You control the denominator. (๑˃ᴗ˂)⁠ﻭ