Most CEOs fire people and blame the market. Jack Dorsey fired 4,000 people and blamed artificial intelligence.
Block, the company behind Square and CashApp, is cutting over 4,000 employees, roughly 40% of its entire workforce. And Dorsey did something unusual: he said why, explicitly, in plain language. AI tools enable smaller, more efficient teams.
The stock went up. Of course it did.
Let me be provocative for a second
This is also marketing. Dorsey said something provocative because provocative statements generate clicks, headlines, and relevance. You need to stay in the news, you need people talking about you. Saying “we restructured for efficiency” gets you a paragraph in a trade publication. Saying “AI can do their jobs” gets you the front page of everything.
There are multiple layers here, and the surface-level story is not the interesting one.
What nobody is saying: it is not intelligence
Here is the thing that bothers me about how this story is being covered. Everyone frames it as “AI is now smart enough to replace humans”. That is not what is happening.
What is actually happening is agentic automation. Procedures being automated. People spend enormous amounts of time on repetitive tasks that follow established procedures, and once you have a procedure, it is just following a recipe. There is no intelligence needed. You do not need artificial intelligence to follow a recipe. You need a system that can read the recipe, execute the steps, and handle minor deviations.
Current AI is extremely shallow. It is the illusion of intelligence, brittle and curated. The human still has the brain and needs to micromanage the system. But if the task is procedural, if it follows a script, then you have eliminated the need for intelligence entirely. A well-built agentic system can follow procedures, scour the internet for information, read manuals, and process data, all in the blink of an eye. And machines do not get tired, do not have bad days, do not have egos.
That is not intelligence replacing people. That is automation replacing procedures. There is a massive difference, and nobody in the press is making it.
The headcount problem nobody talks about
Here is what is unsaid in every layoff announcement: headcount is not always productive.
People have bad days and good days. They have egos, personal interests, and a natural tendency to do the minimum required. They get sick, they go on holiday, they need motivation and management and HR to manage the HR. The bigger the organization, the harder it is to align everyone, the harder it is to change course, the harder it is to even detect when things are going wrong.
Before a year ago, you did not have options. If you needed something done, you hired people. If you could not hire the best people, you hired more people to approximate what the best could do. But humans fluctuate. They hit their stride and then they dip. Everything influences their output: mood, health, relationships, whether they slept well.
I am human. I am not against humans. I am stating facts. And in a competitive, capitalist system where your business depends on selling goods, you either compete on quality, price, or because you are the only option. Headcount makes all three harder.
What I know from experience
I use AI every day. I have zero employees. And I can do the work that would have required 10 to 12 knowledgeable, aligned, focused people. Even then, they probably would not match the output, because alignment and focus are the hardest things to maintain across a team.
But here is what nobody tells you: the hardest part is not the automation. The hardest part is creating the harness, the tooling, the knowledge base, the procedures that make automation possible. I have invested enormous time in building those tools. Without that investment, without being on the bleeding edge and constantly updating myself, none of this works.
The tools are not intelligent. I stress this. There is absolutely zero intelligence in the current systems, not in the way your brain handles it naturally. What looks intelligent to you as a user is curated, managed, and micromanaged by the human who built the harness. Simple things that come naturally to a human brain are not yet defined in the AI’s universe. They are completely oblivious to them.
The same goes for vibe coding, by the way. On the surface it looks incredible: “I just wrote a program that does amazing things without understanding a line of code”. But scratch under the surface and everything crumbles. That is where we are right now.
The part that excites me and terrifies me simultaneously
Not everything can be automated. And the things that cannot be automated will gain more value, not less. Craftsmanship. Intuition. Understanding where trends are heading. The ability to see around corners.
These become premium skills in a world where procedures are free. When everyone can automate the routine, the human who brings something that cannot be scripted becomes the most valuable person in the room.
This is not the first time humanity has been through this. Steam replaced muscle. Electricity replaced steam. Computers replaced calculators. Every revolution displaced jobs and created new categories that did not exist before. This one will too.
The timing matters. Too early and you burn money on tools that are not ready. Too late and your competitors already automated what you are still doing manually. The sweet spot is narrow, and the companies that find it will reap the benefits.
It is exciting, it is scary, and it is happening whether any of us like it or not. The only question is whether you use these tools to build something meaningful or just to cut costs and spin a story about it.
Source: Digital Applied