Will AI replace you? 5 groups that, according to the creator of Claude Code, will survive the AI ​​era. Quiz: Which one are you in?

Over 58% of Poles believe that AI will take away many people’s jobs. The NASK report from 2025 indicates 5 million positions susceptible to automation. Hays Poland reports that 48% of Polish specialists are afraid of limiting opportunities – only in Hungary there are more pessimists.

I understand this fear. Except we usually ask the wrong question.

For the past two years, I have watched the discussion about “AI and work” revolve around one thing: will AI replace me? Meanwhile, the man who created the tool at the center of this discussion suggests a completely different starting point.

‘ + ” + ‘‘ + scores(k) + ‘‘; scoresDiv.appendChild(row); }); } show(‘bpwq-result’); setTimeout(function () { document.querySelectorAll(‘.bpwq-score-bar’).forEach(function (el) { el.style.width = el.getAttribute(‘data-pct’) + ‘%’; }); }, 80); } var startBtn = document.getElementById(‘bpwq-start-btn’); if (startBtn) startBtn.onclick = start; var resetBtn = document.getElementById(‘bpwq-reset-btn’); if (resetBtn) resetBtn.onclick = function () { show(‘bpwq-intro’); }; })();

If the quiz does not load properly, refresh the page.

Who is Boris Cherny and why do I listen to him?

Cherny created Claude Code – the Anthropic coding agent, which started as his side project in September 2024. Today it is used by hundreds of thousands of engineers around the world. When I write about how work is changing, I take it seriously.

A few weeks ago, he posted a thread on X that collected tens of thousands of likes. Not about how to use AI. About how the meaning of work as such is changing.

His observation: the traditional division into engineer, product manager and designer is dissolving. AI equalizes access to technical and design skills. When anyone can write code and create interfaces, something else matters – in which phase of product development you give your best.

Cherny observed his team and identified five archetypes.

Five employee archetypes in the AI ​​era

Prototyper generates ideas. Lots of ideas. Most won’t make it to production – and they know it. The value of Prototyper lies in the speed of the experiment, not in its survivability. In the context of crypto: this is a person who submits a proof-of-concept of a new DeFi mechanism over the weekend. Not everything works. A few things go further. What AI cannot replace here is intuition – which is worth checking.

Builder takes a prototype and brings it to a state where it can be shown to the world. Translates the experiment into a product with scalability and security in mind. I believe this is the most common role in the Polish tech ecosystem: software houses, outsourcing, startups. Builder equipped with AI works many times faster. A Builder without his own opinions about what he builds is becoming more and more easily replaceable.

Sweeper comes in after Builder and cleans up. It simplifies the code, improves UX, and removes technical debt. It doesn’t add features – it removes anything that shouldn’t be there. An underestimated role because its effects are invisible to the user. In the Bitcoin ecosystem, where stability and code minimalism are important, Sweeper has exceptional value.

Grower takes a working product and asks: why do people use it? Why do they stop? What should we change to attract more relevant users? This is the archetype that is missing the most in Polish Web3 projects. The Polish blockchain scene has many Builders and Prototypers – projects are technically sound, but then they die because no one took care of growth, retention or onboarding. A grower doesn’t have to be an engineer. Must understand human behavior and the data correlated with it.

Maintainer it keeps an eye on the system that is already working. Uptime, security, efficiency as the product grows. He doesn’t add new things – he makes sure that the foundation remains unchanged. In the Bitcoin sense: node operators, people managing Lightning infrastructure, custody. Bitcoin as a protocol deliberately favors Maintainers – it changes slowly, because stability is a value in itself. And it is this role that AI finds most difficult. Maintainer understands the system in its concrete, production context – something that the language model cannot reproduce from the description.

One archetype – maybe two

Cherny points out: most people naturally fit into one role, sometimes two adjacent ones. Trying to be everything ends up being mediocre at everything.

Importantly: the right mix of archetypes changes with the product phase. A new project needs a Prototyper and a Builder. Growing – Builder, Sweeper and Grower. Mature system – Sweeper, Grower and Maintainer, and Builder is borrowed for specific tasks. If you are a Maintainer in a project in the experimental stage, the problem is not you. The problem is that you are at the wrong stage for your natural mode.

The right question

Not: “Will AI replace me?”

Or rather: in which phase of product creation do you give your best? Where do you have a natural acceleration and where do you have to force yourself?

Cherny does not provide an answer – he gives a tool for thinking about work differently than through the prism of position and technology. In my opinion, this is one of the better compasses in this discussion, because it does not depend on what the AI ​​tool you are using can do. It is based on the phases of the production cycle, which is what most AI tools look like.

Decide which archetype you want to master.