@MMMusol: In the AI Era, It's Not Just About 'Engineers' Anymore; We Need These Five Types of People. Anthropic Doesn't Only Hire for Frontend or Backend Anymore; It Looks at Which 'Cognitive Archetype' You Belong To. Claude Code Lead Boris Cherny Put It This Way: As Engineering, Product, Design, and Data Science Gradually Merge into New...

The AI era doesn’t just need “engineers” anymore — it needs these five types of people

Anthropic doesn’t just look at front-end or back-end when hiring. It looks at which “cognitive prototype” you belong to.

Here’s what Boris Cherny, head of Claude Code, had to say:

As engineering, product, design, and data science gradually merge into new types of roles, I’ve been thinking about what positions will look like in the future. Looking at the Claude Code team, I see five archetypes:

1. Prototyper: Comes up with brand new ideas, churns out lots of creative concepts quickly — most of which never ship.

2. Builder: Rapidly turns prototypes into production-grade products/infrastructure.

3. Sweeper: Cleans up UI, simplifies systems, retires redundant features, optimizes performance.

4. Grower: Takes over built products and iterates continuously to improve product-market fit (PMF).

5. Maintainer: Owns mature systems, ensuring they remain secure, reliable, fast, and efficient while scaling.

Many people cover two or three roles simultaneously. These roles aren’t tied to job titles — at Anthropic, some designers are Type 1, others are Type 3; the same goes for engineers, PMs, and data scientists.

A healthy team needs the right mix based on product stage:

• Pre-PMF: need 1+2+3
• Growth stage: need 2+3+4+partial 5
• Strong PMF: need 3+4+5+partial 2

“Sweeper” is painfully real, but why separate it from “Maintainer”?

I think Sweeper is an accurate critique of the industry: every team knows tech debt is piling up, but the people actually willing to kill features and delete code are always a minority. AI accelerates code production, which also accelerates garbage accumulation — making Sweepers even scarcer.

My initial confusion: aren’t these both just “code governance”?

The difference lies in timescale and mental model:

• Sweeper is a surgeon — cuts in, removes necrotic tissue, stitches up, and leaves. It’s episodic and requires aesthetic judgment; essentially a design decision.
• Maintainer is a gardener — waters and prunes daily, observes which leaves are yellowing, iterates gradually. It requires deep memory of the system’s full picture.

One needs taste and decisiveness; the other needs patience and systems thinking. “Knowing what shouldn’t exist” and “ensuring what exists stays healthy” — these two temperaments rarely coexist in one person.

So who do top teams hire in the AI era?

Boris gave three criteria at the Fortune Brainstorm Tech conference:

Generalists

“We like people with cognition in more than one area — engineering plus design, engineering plus product, data science plus design.”

As role boundaries dissolve, a Prototyper who doesn’t understand design can’t validate user intuition; a Builder who doesn’t understand product can only build demos in a vacuum.

Low ego

“You have to be able to accept that your own ideas might be wrong.”

Most Prototyper ideas never ship. Sweepers have to cut others’ — and their own — code. Growers have to accept data showing users don’t want it. Maintainers have to accept their work will never make the release notes. Every role requires letting go of ego.

Empiricists

“Learn from data; anchor yourself in reality.”

PMF isn’t imagined — it’s measured. Empiricism means being willing to abandon a belief you’ve invested three months of effort in when the evidence says otherwise.

The common thread: top AI-era talent isn’t about being a full-stack engineer; it’s about being able to switch cognitive dimensions, maintain skepticism toward your own judgments, and make decisions based on data.

But these roles are cognitive stances, not job titles

The sharpest part of Boris’s framework isn’t listing five types — it’s that they aren’t tied to job titles.

The future team-building logic will shift from “two front-end, one back-end, half a PM” to “two Builders, one Sweeper, one Prototyper — background doesn’t matter.”

So for individuals, rather than first asking what to study, ask “what role am I naturally?”

Tech stacks may become obsolete, and AI can write code for you. But “seeing chaos and itching to clean it up” or “getting a half-finished product and wanting to find users” — these instincts can’t be replaced.

When AI generates a complete prototype in minutes, the Prototyper’s value is “knowing which one to build.” When AI auto-refactors code, the Sweeper’s value is “knowing which complexity is unnecessary.” Every role’s value moves from execution layer to judgment layer — and judgment requires cross-domain cognition, low ego, and an empiricist mindset to cultivate.

In an era where AI does the work for you, your job isn’t to work — it’s to decide what should be done.

Boris Cherny (@bcherny): As engineering, product, design, DS, etc. melt into a new kind of role, I was reflecting on what roles might look like in the future. For example, when I look at the Claude Code team I see what I think is five archetypes:

1. Prototyper: comes up with brand new ideas; churns out