Testing For 2019

If you're still banning AI in your take-home, you're testing for 2019.

When you submit our take-home, I can tell it's written by AI. Of course I can. We ask you to use AI. 🙂

That's not what we look at. We look at how you use the AI, how good your quality of output is, and how good the agentic setup in the task is.

I want to know how you spend your time. How do you do quality assurance? What part of your quality assurance is manual testing? Just automatically running a few tests that AI generated is not really testing it. That's the agent grading its own homework. 🫠 It doesn't make sure it's actually solving a real user problem.

For that, you still need human taste. Then there are decisions you make, like UI and UX, that carry a lot of weight. The agents are quite bad at those, and they require a lot of human input and taste. Nobody cares how clean your code is if nobody finishes the flow.

So far, very few candidates show that distinct taste for good quality output. They just accept whatever the agent gives them. And that's how you get mediocre results. 10 years of experience and you still accept the first thing the agent gives you? That's not senior.

Human taste is severely lacking. Product thinking is severely lacking.

Writing code was never the job. Now there's no hiding it.

In this economy, you need to be stepping up your product game more than your engineering game. 💀