A Death Rattle

The most hardcore PMs in my circle have lost their damn minds lately.

They’re all messing around with Claude Code and OpenClaw — stuff that’s still pure command line. To even use these things, some of them fought through nightmare environment setups, some borrowed U.S. credit cards, some literally flew to another country — all to get their hands on it day one.

Watching them stare at green-on-black terminals, fumbling through commands just to crank out a demo — you could feel the sweet revenge dripping off every keystroke: “Holy shit, this is incredible. I never have to deal with those engineers again.”

Everyone thought the sun had finally risen. AI in hand, we’d gone from “the guy who files tickets” to “God.”

I’ll be honest — I was high on it too.

Two weeks ago I used Claude Code to blast through months of backlogged Linear tickets that were never going to get scheduled — and submitted dozens of PRs myself. That rush of “my will, executed directly in code” had me screaming like a kid: “I’m back, baby.”

I mean, I wrote code for eighteen years straight — from winning national CS competitions in ’94 to going PM in 2012. In the decade-plus since, I hadn’t touched a line of real code.

But once the high faded, something started to feel very wrong.

This isn’t a rebirth. It’s a death rattle.

The Last Tiger

If you’ve played Battlefield V, you might remember a single-player chapter called “The Last Tiger.”

Spring 1945, the Battle of the Rhine. You play as a Tiger tank crew that fought all the way from North Africa to the German homeland. Battle-hardened veterans, every one of them. The tank is the most terrifying thing on the battlefield — thick armor, an 88mm gun that could kill anything.

But the war is already lost.

The Tiger used to be king — one could take ten Allied Shermans. By 1945, the Allies had more tanks than you could count, pouring over every hill. Germany was running on fumes. Commander Müller stared at the rubble and kept following orders that meant nothing anymore.

The best equipment, the most experienced crew. None of it mattered.

The Tiger wasn’t weak. The game had changed. Wars aren’t won by how good your soldiers are. They’re won by how many factories you have.

Looking at this AI wave, I have a sinking feeling: we PMs are that Tiger tank.

The Wisdom of Poverty

I’ve had a good run. Led big teams. Shipped more hit products than I can count on both hands. I’m still proud of what I do: I’m great at trade-offs, great at strategy, great at saying no.

Those skills used to put me on top.

Then I watched AI spit out code in seconds, and a brutal truth hit me:

Everything we worship — “strategic thinking,” “MVP,” “prioritization,” “stay lean and disciplined” — none of it was wisdom. It was poverty.

Just us making do because code cost too damn much.

Engineering was expensive. Programmers were scarce. So I couldn’t build everything — had to cut. Couldn’t serve everyone — had to pick the core users.

But what if code is free? What if building stuff costs nothing?

Who’s going to sit there agonizing over P0 vs. P1? Who’s going to debate Plan A vs. Plan B? Screw it — build both, try everything. That’s what people actually want.

When trial-and-error costs nothing, my precious “strategic discipline” is a punchline. My “prioritization skills” are just survival tricks from the age of scarcity.

Like the Tiger’s heavy armor and 88mm gun — unbeatable when there’s fuel. But when you can’t even fill the tank, all that weight is just dead weight.

The Greatest Common Denominator

Because code was expensive, we had to find a “greatest common denominator.”

That GCD is today’s app — a standardized product everyone can use but nobody fully loves.

A product manager is, at bottom, the person who computes that greatest common denominator.

I think about my career and most days I wasn’t chasing the ideal experience. I was hunting for the best compromise I could get on a tight budget.

Even ByteDance — mighty ByteDance — nailed personalized content feeds, but the app UI? Same for everyone. Why? Because before AI blew up, personalizing software at the architecture level cost a fortune.

As long as code is expensive, companies need someone like me to find the “standard answer.”

That premise is now falling apart.

Death of the Middleman

People say I’m being dramatic. PMs are still the ones running projects and wrangling teams, right?

Think about what happens next.

Next 0–6 months: PMs get AI tools and can validate ideas fast. Looks like a superpower. Actually it’s the first crack in our leverage. Before, the boss had an idea and had to come to me: is this even feasible? Now he just says: “Why don’t you mock it up with AI and show me?”

I went from gatekeeper to demo monkey.

6–12 months: Execs and business people talk to AI directly and get product prototypes. They don’t need us to “translate requirements” anymore.

You think our value was “understanding the user”? Nope. It was “turning requirements into something engineers could parse.”

Like human translators — the real moat was never “knowing two languages.” It was the switching cost. AI kills the switching cost. The middle layer evaporates. We become noise.

Beyond one year: This is where it gets scary. When generation costs hit zero, the whole idea of a “greatest common denominator” collapses.

I want a fitness app with just the check-in features I care about, in a cyberpunk skin I like. AI builds it on the spot. Use it, toss it.

Your company’s CRM isn’t the off-the-shelf version IT bought anymore — every sales rep gets their own custom build.

Games stop designing difficulty for “the average player” — every player’s entire experience is generated in real time.

In this world, the “standard product” dies. No common denominator, no need for a calculator.

Don’t Bet on the Tech Plateau

I know the pushback: “Have you tried deploying OpenClaw? Normal people can’t even set it up.”

Yeah, the tools are rough right now. The bar is high. But don’t look at AI and assume it stays this way.

ChatGPT has existed for barely two years. Clunky tools are a temporary problem — and the window is shorter than you think.

Betting that the tech will plateau? That’s hubris.

In 1944 the Germans still thought the Tiger could turn things around. Spring of ’45, they were still mounting counterattacks on the Rhine.

Meanwhile one American factory was pumping out dozens of Shermans a day. Escort carriers rolling off the assembly line like dumplings into a pot. And Germany couldn’t scrape together enough gas to move.

You can sometimes close a technology gap. You cannot close a production gap. That’s a steamroller.

Endgame

The “empathy” and “user insight” we’re so proud of? Just us paying down the debt of expensive engineering.

The debt is paid. And the creditor just got a direct line to the money printer.

The Industrial Revolution killed the coachman because cars don’t eat hay. The AI revolution will kill the product manager because AI doesn’t need a middleman.

This is how it ends.

In the final scene of “The Last Tiger,” nearly the whole crew is dead. Müller sees it for what it is. He surrenders. Throws away his Iron Cross. But Schroeder — the young gunner, brainwashed to the bone — can’t accept it. He picks up an MP40 and points it at Müller.

Smash cut to black. One gunshot.

Müller fought from North Africa to the Rhine. Dodged a million enemy rounds. But he couldn’t dodge a bullet from his own crew.

He wasn’t too weak. When the system starts to collapse, the first one sacrificed is usually the first one who saw it coming.

I watch AI-generated code scroll down my screen. I know exactly how Müller felt.