"AI allows engineers and designers to do everything alone."
"Specialized professions are no longer needed."
These kinds of statements have been increasing lately.
AI has indeed caused individual productivity to explode. An era has arrived where one person can play many roles.
However,
While we hear that "one person can do everything," it's not as if the world is overflowing with people who are actually imitating that and achieving high quality. It's only a tiny fraction. Many people cannot do it even if they try.
Perhaps everyone is lost.
In the midst of this, the man who created "Claude Code" is saying the exact opposite.
Observations from the Man Who Created Claude Code
Boris Cherny.
The Anthropic engineer who led the development of Claude Code.
In other words, he is someone who has witnessed at the world's front lines "how AI changes people's work."
Boris observed his own team and said this:
People are divided into five "types."
The 5 Types
https://x.com/bcherny/status/2071379474277613732
・Prototyper (The Originator)
Generates new ideas one after another. Creates in bulk, most of which are shelved. The 0→1 person.
The type who brings up ideas in meetings that no one else thought of, asking, "Isn't this interesting?" They compete on the volume of ideas rather than perfection.
・Builder (The Constructor)
Takes a prototype and brings it up to a usable product level in one go. Speed and implementation power.
They turn ideas that most people just call "interesting" into something working by the following week. The person who converts the Prototyper's ideas into reality.
・Sweeper (The Refiner)
Polishes the UI, organizes code, and cuts unnecessary parts. Someone who can make "subtraction" judgments.
After release, they steadily fix things like "this is confusing" or "this feature isn't needed." A type that increases value by removing rather than adding. Loves practical work.
・Grower (The Scaler)
Grows a functioning product. Increases product-market fit; the protagonist after PMF.
While looking at numbers, they repeat tasks like "changing this flow will reduce churn." They aren't flashy, but they are usually the ones who double the revenue.
・Maintainer (The Protector)
Ensures the safety, reliability, and speed of a mature product. Builds a foundation that doesn't break even when scaled.
The person who ensures the system doesn't crash even with a million users. Creating a state where "nothing happens" is the achievement of this type.
Your Title Doesn't Explain Your Strength
This is the core of Boris's point. These five types do not match job titles.
Among "engineers," some are Prototypers and others are Maintainers.
Among "designers," some are Builder types and others are Sweeper types.
The same goes for PMs and data scientists. Even if the title is the same, the pattern of value they provide is different.
→ Defining yourself by a profession like "I am an engineer" or "I am a designer" explains nothing about your essential strengths.
The Trap of "Doing Everything Alone"
Some people might wonder:
"In the end, isn't a generalist the strongest?"
The range one person can cover with AI has certainly expanded.
However, what Boris found by observing his team was the opposite fact. Even the Claude Code team, which masters AI the most, has these five types.
Why?
The judgment required of a Prototyper is: "Is it worth trying something no one has tried yet?"
The judgment required of a Maintainer is: "Will this change break the entire system?"
These two are opposites.
The former creates value by taking risks, while the latter creates value by eliminating risks. AI makes the work of both faster. However, the thinking habit of which judgment to prioritize cannot be changed.
Doing everything alone means the same person is stepping on the accelerator and the brake at the same time. It's not impossible, but both will be half-baked.
A team of people who score 120 in one type is stronger than a generalist who does everything at an 80-point level.
Boris himself clearly states, "A healthy team needs a mix of types."
AI raises the floor of productivity for each type. The difference between the types themselves does not disappear.
The Required Type Changes with the Phase
Boris also points out that the necessary combination changes depending on the product phase.
・Launch Phase (Pre-PMF): Prototyper + Builder + Sweeper. Create, try, and polish.
・Growth Phase (Post-PMF): Builder + Sweeper + Grower. Stabilize while growing.
・Maturity Phase: Sweeper + Grower + Maintainer. Keep improving while protecting.
This isn't limited to companies. The same dynamics exist in sales organizations, planning departments, and back offices.
This is because every job has phases of "creating," "shaping," "organizing," "growing," and "protecting."
Value That Drops vs. Value That Doesn't in the AI Era
People who are aware of their type can choose the environment where they can provide the most value.
Conversely, people who aren't aware of their type choose environments based on "titles." That means entrusting your market value to a title.
In the AI era, the value of titles will drop. The value of types will not.
This is because AI replaces "tasks," not "types."
In the AI era, instead of trying to do everything yourself, master a specific type. That will form your career.
Those who master one thing will remain, rather than those who do everything.
If even one of your types is put into words, your career axis has already started moving.
