Anthropic's Hit List, the End of Body Shopping, and My Orchestration System

In the tech industry of Latin America and the Río de la Plata region, for years we got used to taking a slice of development work from the United States by leveraging our time zone and competitive costs.

Today, the equation has partially flipped: much of the transactional work is fleeing to India for their rock-bottom rates. But the cycle closes perfectly: in less than five years, Artificial Intelligence will take that work from them.

That's a joke (for those who missed it), but if you look at the raw corporate adoption numbers in the region, it's a mathematical prophecy. The historical model of selling basic programming man-hours — the famous body shopping — is collapsing under its own weight.

And if you think this is cheap futurology, look at what Anthropic published this week. Basically, the people building the guillotine had the decency to publish the exact list of who's going to lose their head. A level of transparency that is as beautiful as it is terrifying.

The raw facts without anesthesia:

• Ground zero: Programmers, financial analysts, and customer service representatives are today the most exposed roles on the planet. 75% of a programmer's tasks can already be automated by Claude. Today. Not in 2030.
• The university lie: Who's most threatened? The older workers, with high salaries and university degrees. Exactly the demographic we told: "get a degree and you'll be safe."
• The invisible wall: Junior hiring in these roles has dropped 14% since ChatGPT launched. Companies aren't laying off en masse; they simply stopped hiring. The class of 2026 is walking toward a wall they haven't seen yet.
• The physical lifeline: The safest jobs? Bartenders, dishwashers, and lifeguards. 30% of the labor market is protected by physical reality, not credentials.

The darkest part of the entire report is the gap between what AI can do and what corporations are currently using. AI is already capable of automating nearly all knowledge work. The only thing slowing it down is corporate inertia, legal bureaucracy, and fear. This isn't a market skills problem; it's purely an adoption curve.

And here's my reading: this isn't a US problem — it's a global checkmate. India has 5 million IT workers in the line of fire. The Philippines has 1.5 million in call centers and BPO. Entire economies built on knowledge work arbitrage, and AI just eliminated their only competitive advantage overnight. LatAm can see itself in that mirror.

Titles and certifications aren't dead, but they're no longer enough. The competence signal mutated: we moved from credentials to raw capability. From diplomas to demonstrations. From what you studied, to what you can ship to production.

The Slot Machine Trap

The problem is that most people are facing this transition in the worst possible way. Professionals treat AI like a casino slot machine. They put in a one-line generic prompt, pull the lever, and pray for perfect code. When AI returns mediocrity, they get frustrated and go back to their ten-year-old manual processes.

AI demands that you be an architect, not a gambler. It requires a structured workflow.

Proof of Work: My System in the Trenches

To make it concrete: my new personal site, pablomarichal.com. Running in production, with all code audited on GitHub.

I have plenty of experience with WordPress; I built and sold several projects of this type in my pre-SAP days. Last year I lost money and sanity buying Elementor licenses, wrestling with heavy plugins and WP's bloated code. I built something I ended up throwing in the trash.

This week I changed the rules. I developed and shipped to production the site in a few real hours of maximum effort.

The stack: Claude Code directly in my terminal as the execution engine, complemented with Antigravity. When Anthropic's model left me out of tokens from the daily limit, I pivoted to Gemini Pro 3.1 and kept pushing the architecture without losing a second of momentum.

It's an absolute pleasure watching those two engines run at full power while I simply orchestrate.

There's a reason the industry already says writing code by hand is a thing of the past. There will be syntax craftspeople, but they'll do it with the same romanticism as those who knit scarves or brew craft beer in the garage. You won't scale a business that way.

Doing things well pays immediate dividends: operating with AI isn't just throwing a prompt. It's knowing how to connect APIs, define the user experience, and maintain absolute control over the logic. AI coded, but the product direction was mine.

The value is no longer in being the bricklayer who lays the bricks — it's in being the orchestrator who designs the building.

The Leverage

If you want to understand where the world of automation is headed, read The Almanack of Naval Ravikant (free online). Naval explains better than anyone why code and content creation are the definitive forms of modern leverage with no marginal cost.

The signal of competence has mutated. We moved from credentials to raw capability. From what you studied, to what you can demonstrate and ship.

The market doesn't wait for anyone. Build while others sleep.

Pablo / En Producción