In his conversation with Kent C. Dodds, Gabriel Benmergui heard a blunt warning for developers who want to keep typing all their code themselves: “you've got to wake up.” Kent's concern is the time they lose getting useful changes into users' hands.
For a founder hiring engineers in Latin America, that makes AI adoption worth examining before you make an offer. You need someone who can ship with agents and still explain why the result is safe to release. A candidate's attachment to doing the typing themselves deserves as much scrutiny as their enthusiasm for generating code.
The cost of insisting on the keyboard
Kent starts by describing developers who care about code and want to keep writing it by hand. He anticipates their objection because that preference is part of what he wants them to reconsider. His warning is direct:
“Um, you cannot be hand typing the code anymore.”
Later, he puts the consequence in business terms:
“And you will just get outpaced by your competitors.”
Ask an engineer to walk through a recent change they shipped with a coding agent. Listen for how they supplied context and checked the output. You can learn more from that account than from a list of tools on a résumé.
If they prefer doing a particular task by hand, ask why. There may be a good reason: perhaps understanding the existing behavior took longer than writing the fix. You're looking for a deliberate choice about how to finish the work, rather than a refusal to change how they work.
Quality is part of the argument
Kent explicitly rejects speed as the only thing that matters:
“It certainly is not. And quality product is what matters.”
That qualification belongs next to the warning, especially when you use the clip to think about hiring. A developer who accepts an agent's output without understanding it hasn't demonstrated the judgment you need. Someone still has to decide whether the change addresses the user's problem and whether the team can maintain it.
The engineer remains responsible for the result. During an interview, stay with the candidate's example long enough to hear how they reviewed the diff. Ask what they tested and what they rejected. If the agent produced a plausible but incorrect answer, have them explain how they caught it.
These are useful engineering questions even when the candidate uses no AI. Coding agents make the answers more important because a person can now produce changes they haven't taken the time to understand.
Look at the work users are waiting for
Kent names bug fixes and feature requests as places where speed matters. Those are familiar hiring needs for a startup: you already have users asking for something, and you need an engineer who can deliver it.
Evaluate the whole path to release. Generating an implementation quickly helps little if another engineer has to spend the afternoon untangling it. Ask candidates how they keep changes reviewable and what evidence they give a teammate before asking for approval. Their answer should make the work easier to inspect.
The other cuts from this interview add useful context. In Coding agents can't find the market, we discuss why building software still requires feedback from users. In Your next customer is an agent, Kent considers who will use that software. Read them alongside this clip: faster implementation leaves founders with product decisions to make and engineers with responsibility for what they ship.
Bring AI adoption into the interview
At Silver.dev, we evaluate how engineers work with agents in our AI adoption interviews. The AI adoption kickstart and the AI adoption interview guide, both in Spanish, show candidates how we approach that work.
For your next hire, ask for an example you can examine together. Give the candidate room to explain a mistake or a tradeoff. An engineer who can identify where an agent went wrong and show how they verified the correction gives you something concrete to assess.
If you're hiring in Latin America, talk to Silver.dev about engineers who use agents and own their work. Watch the clip with its transcript here, or return to Gabriel's full conversation with Kent C. Dodds.