You ever notice how every few decades we announce the future, and then we ignore it for ten years out of spite.

Like, right now, we have these AI systems that can write code, test it, deploy it, document it, and then politely ask if you would like a second application with fries. And the response from a lot of companies is, “Interesting. Anyway, let’s do standup at 9:15 and then argue about Jira for forty five minutes.”

It is a fascinating moment in human history because the tool is sitting there, doing the thing, and we are still acting like the real work is the ceremony around the thing. Leading companies no longer have developers writing code, instead orchestrating projects that do the work of dozens. But for now most other companies kept the priesthood, robes, and incense to prove their faith in the old and obviously obsolete way of doing things. After inviting a machine that can build a cathedral in an afternoon they said, “Great. Now please wait while we align stakeholders.”

The funniest part is how dignified we try to be about it.

A guy will tell you he is a “Python developer.” Like that is his identity is being born in a Python nest and raised by snakes, and you are supposed to treat this with reverence. Meanwhile the AI is cranking out the same code in ten seconds, plus unit tests, plus integration tests, plus a little note at the bottom that says, “I also fixed your typo. You are welcome.”

So what do we do? We cling to comfortable old rituals despite seeing they have no future. We protect the old job shapes and little laminated badges.

We say, “We need a team.” Why? The “team” used to be how you got work done. Now the work gets done by a system that does not need a break, does not need encouragement, and does not suddenly decide it wants to rewrite everything in Rust because it watched a YouTube video. You can feel the panic coming off the old process like heat from a radiator.

Traditional development that people learned in the era of keyboards and human ability was built around scarce attention, time, coordination, and competence. So we invented all these structures to manage the scarcity using meetings, ticket queues, and approval. Entire careers were built around moving information from one inbox to another.

Now an agent can plan, execute, verify, and revise. It can hold the context in its head without getting tired. It can remember what you said last week without pretending it forgot so it can argue again. It makes the old setup look like a horse and buggy on the highway.

The horse is doing his best, sweating as it works diligently. It’s a noble, dignified creature we all love. The horse is also going twelve miles an hour and you are trying to merge into traffic with people doing seventy five while eating a breakfast sandwich and yelling at Siri.

So you get this weird spectacle where a company installs the future and then immediately handcuffs it. They buy the tools, brag about their AI initiative, and begin doing pilots. They make a slide deck. They hire a “Head of AI Transformation.” Then they instruct everyone to keep doing everything the old way because change is “risky.”

Risky compared to what? Compared to having to change to the reality of being outcompeted by a smaller group with better leverage.

Agentic development changes leverage so hard it becomes almost rude. One person, with the right tooling, can behave like an entire product org. You describe what you want, the agent drafts the architecture, writes the code, wires the database, builds the interface, generates the tests, spins up the environment, checks the edge cases, and then asks if the copy on the button should be “Submit” or “Continue.”

And the old crew is still holding a meeting to decide whether they should create a spike to evaluate the idea of a button.

There is something comical about a world where the main obstacle is willingness rather than ability. The constraints are cultural and emotional. It is the terror of admitting that the skills you built your identity around have been turned into a commodity. To avoid reaching the near-term conclusions, you see a lot of people reach for comforting phrases.

They say, “AI will assist developers.” Sure. It will assist them the way a bulldozer assists a guy with a shovel. The guy is still there. He is still involved and contributing in a sense. He is also irrelevant to the outcome.

They say, “AI cannot understand requirements.” It can, if the requirements are clear. The problem is the humans rarely are. The agent does not fail because it is dumb. It fails because we like to speak in fog. We call it “flexibility.” It is actually indecision and internal confusion, which always leads to failure in every aspect of life.

They say, “We still need governance.” Yes, you always need governance. The difference is that the governance becomes the job. The job is specifying the intent to capture what matters, what is allowed, what must be true when the system is done, and what tradeoffs are acceptable.

That is an uncomfortable shift because it forces honesty. If the agent can build anything, then your real limitation is what you can think through. You cannot hide behind process anymore. You cannot say, “We would have built it, but we did not have the sprint capacity.” The sprint capacity is infinite. The clarity is scarce. Every excuse looks as ridiculous as holding on tight to antiquated technology and practices.

Traditional development becomes obsolete in the same way the elevator operator became obsolete. The operator disappeared because the button was better at being a button. Why would keep a developer who is a hundred times slower, needs a team of people, lacks knowledge in several important project aspects, and isn’t cost effective?

And this is where the comedy turns slightly dark, because the same logic applies to a lot of the white collar stack. Once you can orchestrate work, review work, test work, and summarize work with software, then all the roles built around coordinating humans start to look like decorative hats.

Some people will adapt. They will become the people who can aim the system. They will become high leverage designers of intent. They will get very good at decomposing reality into constraints and testable outcomes. They will own the loop.

Some people will not. They will cling to the old motions and insist the motions are the job. They will insist the ceremony is sacred. They will insist it is “best practice” to take six months to deliver what a competent agent could draft by lunchtime.

And those people will become the new typewriter repairmen. Which is not a moral judgment. It is just what happens when the world stops paying for a specific kind of slowness. Some will still call it “innovation” while doing everything possible to make it behave like 2007.

Then, one day, a competitor will quietly stop pretending. They will run the company like the new world is real. They will ship faster, cheaper, and cleaner. They will have fewer people, fewer meetings, fewer excuses. And everyone else will gather around a conference table, acting surprised and blind-sided, and ask how this happened.

And some guy will say, “We should have prepared.”

Which is my favorite line in business. “We should have prepared.” It is like watching a man stand in the rain, soaking wet, staring at a closed umbrella, and saying, “In hindsight, we should have considered precipitation.”