The Day a Mexican Lawyers' Chat Became Silicon Valley

View on LinkedIn

On Tuesday, April 28, at 11:53 in the morning Mexico City time, Claude went down across much of Mexico. In San Francisco, where I live, it was working fine. I was preparing Wednesday's trademarks class, completely unaware, until WhatsApp started buzzing.

It was the lawyers' group.

"Same here. It's down."

"Me too, since this morning."

"I have two accounts and I can't get into either one."

Mathy sent the screenshot: that gray message Silicon Valley has been seeing for months. Claude is temporarily unavailable. We are working on it. Fidel, calling in from Tabasco, asked whether it was just him or everyone. Lu asked a question in Telegram about Manus tokens that only makes sense if you've spent months thinking about these things. Twenty-seven minutes. Nine lawyers online mid-morning on a workday, discussing — not a case, not a client, not an isolated ruling from the Supreme Court — but the possible technical causes of the outage. Whether it was the VPN. Whether it was the signal. Whether tokens were billed separately.

And at 12:20, someone wrote the line that shifted the temperature of the chat: "that's the damn problem with working only with AI… the new generations are going to have their brains atrophy."

There were two laugh emojis. And then silence.

I'm in other chats. Chats of engineers and founders in San Francisco, chats where the people building these models share memes when the infrastructure they themselves maintain goes down. I have several saved. When Claude or ChatGPT go down, those chats are identical: screenshots, jokes, someone asking whether it's their VPN, someone answering that it's not. It's a genre of conversation. I've seen it a hundred times.

What I had never seen, until that Tuesday, was that genre of conversation in Spanish, among Mexican lawyers. I sent the group a meme of stressed engineers when Claude goes down on them. They finally got it. The laughter was one of recognition.

And I was left thinking: something just shifted.

The genre of conversation

There is a type of chat that has existed in Silicon Valley for years. It's the chat of people who build or use frontier software — people whose productivity depends on tools that haven't quite finished existing yet, tools that go down on Tuesday mornings without warning. The culture of those chats is very specific: productive panic, dark humor, instant cooperation to solve the problem, and a question that always floats in the air: if this goes down, how screwed am I?

That culture did not exist in the Mexican legal profession eighteen months ago. It did not exist. If Claude had gone down in October 2024, no lawyers' chat in Mexico would have reacted. No one would have noticed, because no one was using it seriously.

On April 28, 2026, that chat existed. It happened. I lived it from San Francisco, reading it the way I would read an engineers' chat, and for the first time the boundary between the two worlds I inhabit — living in the world where AI is being built, and practicing Mexican law — dissolved for twenty-seven minutes. The lawyers in my country were having the same conversation as the engineers on my street.

That means something. It is not trivial.

The ones who already bought in

The easy intuition about AI in Mexican law is that there's a market left to convince. That more evangelizing is needed. That someone needs to explain to lawyers what this is all about. That intuition is out of date.

There is a group — still a minority, but a real one — that has already moved past that phase. They already bought in. No one needs to explain anything to them. Pedro's line in the chat — "their brains are going to atrophy" — is not the complaint of someone who rejects AI. It is the complaint of someone who uses it, who caught themselves depending on it, and who is unsettled by their own dependence. That discomfort is only felt by someone who is already inside.

The data confirms what the chat was suggesting. Thirty-three percent of corporate legal departments in Mexico already use generative AI, a figure that places the country above the Latin American average. But that data measures the clients — the in-house teams at companies — not the law firms that serve them. And it is in law firms where the asymmetry becomes interesting: globally, firms with more than 500 lawyers have already integrated AI at 100%, while small firms — 1 to 100 lawyers — show 68% that still have not adopted it. The vast majority of the Mexican legal profession fits into that second category.

What struck me about the WhatsApp chat was not Pedro's complaint. It was that nine Mexican lawyers existed online, simultaneously, talking about Manus tokens and Claude screenshots, for twenty-seven straight minutes, during working hours. That is Silicon Valley. That is a founders' chat. And those nine lawyers have no way back. What scared them that Tuesday was not AI. It was losing it. The conversation was not about adoption. It was about withdrawal.

The ones who never noticed

The most interesting thing about April 28 is not what happened in my WhatsApp. It is what did not happen in a thousand other WhatsApp groups.

At that very same hour, in thousands of law firms across the country, Tuesday was unfolding without incident. Lawyers drafting by hand, searching MARCANET one entry at a time, opinions that take two days because they have always taken two days. They did not know Claude was down. They do not know what Claude is. If someone explained it to them, they would say that's for the young crowd.

They are not stupid. They are not lazy. Many of them have been doing their jobs well for thirty or forty years. The problem is not that they are doing something wrong. The problem is that the playing field is shifting under their feet in silence, and the silence is the most dangerous symptom. Those who already use AI have live chats when it goes down. Those who don't use it don't have those chats. They don't have the conversation. They don't have the reflex. They don't have the asymmetry built into their mental routine.

And that difference, today, is not yet fatal. Longstanding clients are still longstanding clients. But the 33% of legal departments already using generative AI, when they renew their outside counsel relationships, are going to compare deliverables. They are probably already starting to compare. The obvious comparison to the notaries of the nineties who didn't want computers falls short — those professionals had fifteen years. This time, that's not on the table.

What I saw from San Francisco

I lived April 28 from an unusual vantage point: with Claude working, in a city where these chats are the daily liturgy, watching my Mexican colleagues have their first collective run-through of the genre. I sent the engineers' meme, the lawyers laughed, and for the first time I felt that the two communities I live in — frontier software and Mexican law — had shared the same Tuesday.

That was not possible eighteen months ago. And in eighteen more months, chats like that are not going to be unusual. They are going to be the standard for the segment of the profession that stays in the game.

Pedro's question — am I too dependent on this? — is the right question, but the short answer is not the important one. The short answer is yes, of course, just as we are all dependent on the digital Semanario Judicial, on the SAT online portal, on the DOF, on Google. Every civilization lives by externalizing capabilities. That question, asked by a Mexican lawyer in 2026, is not a sign of regression. It is a sign of belonging. Only someone who has already crossed to the other side asks it.

The interesting question is the one that is not being asked — the one that is not being asked in the thousands of law firms where Tuesday, April 28, was just another Tuesday. Those firms don't have the chat. They don't have the meme. They don't have the reflex. And that, not Pedro's dependence, is what is going to define who is still seriously practicing Mexican law in 2030.

The issue, as always, is not the tool. It is knowing which conversation you are in. If your WhatsApp chat on Tuesday at 11:53 in the morning looked like a Silicon Valley founder's chat, you are in the game. If it didn't look like anything, if nothing happened, if it was just another Tuesday — that is the question worth asking. Not Pedro's.