Two existential crises in the most fun year of building software

In Gabriel Benmergui's conversation with Kent C. Dodds, one of the first things Kent talked about was how much he is enjoying his work. Then he added the other half:

“AI obviously has changed everything for software engineers, and it's been more fun than I've ever had. Like I've never had so much fun building software, but at the same time, I've already had two existential crises this year”

Kent holds both of those at once. If you hire engineers right now, many of your candidates probably do too, and how they worked through it is worth asking about.

The fun part is real

Kent is no reluctant convert. Later in the conversation he traces his path from pasting code into ChatGPT in late 2022, to tab completion, to running cloud agents in parallel, to a year in which he has barely typed any code. In another clip from the same conversation, he warns engineers who still insist on typing every line by hand that their competitors will outpace them.

So this crisis doesn't come from someone losing to the tools. He uses them as hard as anyone, and he enjoys it.

Where the crisis comes from

Kent is worried about the slope. In the clip he describes “thinking like what is it that I do in this like new world because the trajectory of the AI is just to keep on improving all the time, and eventually is it just gonna completely do away with what I do.”

In the full conversation he explains why he took that seriously. In early 2025 he expected AI to stay mostly tab complete, with him still working inside the code. A year later he wasn't looking at the code at all, and the software was still good. If he couldn't predict that year, he asks, what makes him think he can predict the next two?

For him the question was also about his business. He has made his living from premium courses that take a year or more to prepare, and in a field that shifts every few weeks, a course can be out of date by the time it ships.

Where engineers fit

The clip stops before the answer, but Kent says what he set out to find: “where I landed on that and where I think that engineers fit for as long as anybody fits.”

That last phrase is the honest part. He doesn't promise the job lasts forever. Further into the interview he runs a thought experiment: imagine the last software engineer still doing something valuable, and ask what that person does. His answer is product engineering, which he describes as tying the outcomes a product needs to the technical implementation that delivers them. He also names trust and people's attention as uniquely human, and says that doing uniquely human things is the durable skill.

You can disagree with where he landed. It's still the right question to ask about every engineer on your team, including the ones who are thriving with agents today.

The AI adoption curve

Gabriel's follow-up sets up the rest of the conversation:

“So one of the common concepts here is the AI adoption curve, right? You have the skeptic to the full-blown prompt engineer alone, et cetera. Where would you place yourself in that curve?”

Kent answers in the full conversation and puts himself at the far end of it. That explains the crisis. The closer you are to the front of the curve, the sooner you see what the tools can do, and the sooner you start asking what is left for you.

Every engineer you interview sits somewhere on that curve. A skeptic may be careful and slow to ship. Someone at the far end may ship a lot and still be working out what their job is turning into. Where they sit tells you less than whether they have thought about where it leads.

What to ask when you hire

Kent's two crises translate into questions you can use in an interview:

• Where do you place yourself on the AI adoption curve, and what moved you there?
• Which part of your work do you expect agents to take next, and what are you getting better at instead?
• Walk me through a product outcome you shipped. Which technical decisions made it work?

An engineer who has worked through this will answer with specifics from their own work. One who hasn't will usually talk about tools.

At Silver.dev, our AI adoption interview looks at how engineers actually work with agents. You can read how we prepare candidates in the AI adoption kickstart and how the AI adoption interview works (both in Spanish). For more from the same conversation, read why your next customer is an agent and why coding agents can't find product-market fit.

Watch the full conversation with Kent C. Dodds on YouTube. If you're hiring engineers in Latin America who already work this way, talk to us.