Let’s begin with the truth: Claude is a marvel of modern engineering.

Sonnet 4.5 and Opus aren’t just “good models.” They are, at this moment, the closest thing to a senior software engineer that silicon has ever produced. The logical flow, the context retention, the ability to navigate complex multi-file projects without hallucinating—it’s all so clean, so quiet. It is unlike the noisy chaos we’ve come to expect from LLMs.

Claude feels less like a chatbot and more like stepping into a pristine, soundproof laboratory.

Where others ramble, it reasons. Where others guess, it constructs.

Anthropic’s engineers built a Ferrari engine in a world still driving 1998 Corollas. And then they handed the keys to Marketing. And Marketing bolted a coin slot to the dashboard that shuts off the engine every 15 minutes.

The Engineering–Business Schism

The brilliance of the model stands in violent contrast to the system wrapped around it.

• The Product: A precision instrument capable of real-world problem-solving on a level LLMs have only pretended to reach until now.

• The Delivery System: A paranoid, restrictive experience designed as if the user is a toddler attempting to operate a nuclear reactor.

Developers came for an AI colleague. Instead, they got a paywalled escape room.

Anthropic’s internal conflict is painfully visible: Engineering is building the future, while Marketing is trying to monetize it like a slot machine.

The “Pro” Subscription Betrayal

The user experience is so absurd it borders on satire. Let’s trace the user journey:

1. You hit the limit on the free tier. No problem.

2. You decide the product is worth money.

3. You pay the €22 Pro fee.

4. You expect the shackles to fall off.

Instead… You send a single message and get instantly blocked again.

Why? Because the system counts your free-tier usage from yesterday against your paid-tier quota today. It’s like buying a full tank of gas and being told you can’t drive because last week’s fumes still count against you.

This isn’t a bug. This is a decision.

In any sane SaaS system, payment triggers a resource reset. Anthropic does the opposite: it punishes you for upgrading. A user pays for a new beginning — and gets retrospective imprisonment.

The MBA Virus

This behavior is not something an engineer would design. This is pure business-department pathology. It’s the same disease that has been rotting the tech sector for 20 years.

The MBA Virus: The belief that the best way to maximize revenue is to torture the user experience up to the exact edge where they almost churn.

We’ve seen this play out before:

Windows Phone: Technically magnificent. Strategically murdered.

Tango Gameworks: Delivered a critically acclaimed hit (Hi-Fi Rush). Microsoft immediately shut it down to “realign resources.”

Nokia Hardware: Acquired, drained, discarded.

The Pattern is clear:

1. Engineers create value.

2. Accountants extract value.

3. Executives kill value.

Anthropic is now running this playbook beat-for-beat.

Turning Gold into Lead

The tragedy is not subtle—it’s structural. Anthropic took one of the best LLMs ever built and wrapped it in:

• Opaque usage caps.

• Non-resetting token counters.

• Anti-user billing logic.

• An upsell pipeline that triggers after 45 minutes of heavy coding.

When a single developer hits the “Pro” cap, the system immediately suggests a €90/month Team Plan.

This isn’t scalability. This is extortion dressed as pricing.

You don’t increase ARPU (Average Revenue Per User) by crippling the people who love your product. You only accelerate migration to the first competitor who respects them.

The Tipping Point

 

Anthropic is approaching a dangerous threshold: The moment when the user stops feeling impressed and starts feeling insulted.

Once a developer feels cheated, no amount of model quality will win them back. Trust, once broken, does not regenerate. The engineering team has delivered brilliance, but the business unit is suffocating it in its crib.

• You can’t sell a Ferrari like a gacha game.

• You can’t market a genius like a vending machine.

• You can’t keep your strongest advocates if your system treats them like abusers of your generosity.

Conclusion: A Message to Anthropic

To the Engineers: You built something extraordinary.  To the Business and Marketing Leadership: You are the bottleneck, the liability, and the existential risk to your own company. If you continue degrading the experience in the pursuit of short-term ARPU gains, you will lose the very community that made Anthropic matter in the first place.

Great technology doesn’t die from competition. It dies when the user feels scammed.

Right now, you’re balancing on that edge.

1. Reset the quotas.

2. Fix the billing logic.

3. Treat your users like adults.

Or watch the migration begin.

UPDATE: A Real-Time Postscript

The irony completed itself moments after I drafted this article.

I attempted to generate a short briefing—a simple summary of our session—so I could migrate my work to a new chat window.

The result? Blocked.

A red banner appeared telling me I had exceeded the length limit. I am now locked out of the very context I built, unable to even extract a summary to move forward. The tool locked the door from the inside.

The Final Verdict: Claude’s engineering is one of the finest embodiments of the human mind. It is a triumph of reason. But this service, multiplied by the greed of incompetent marketers, turns a masterpiece into garbage.

 

Follow the Vibe