Solving Problems, Not Puzzles
On AI as the ultimate Rorschach test for software developers
I just read this post from Segment co-founder and former OpenAI employee Calvin French-Owen about vibe coding (or more accurately, AI-powered code agents), and in particular, why some engineers are head-over-heels, and others, well aren’t.
I don’t think I’ve ever seen this divide as clearly articulated as Calvin explains it below.
In this regard, I think AI is the ultimate Rorschach test for engineers.
If you’re passionate about puzzles and writing code for the craft of writing code, of course you’re skeptical of AI writing more of your code. Through this lens, it’s actually a perfectly understandable and rational point of view.
On the other hand, if you’re seriously passionate about solving problems for customers, it’s clear why you would think AI code agents are an amazing innovation. Your ability to solve more problems, more quickly, has increased drastically.
The dilemma, as Calvin raises in his post, is that as AI gets better at “solving puzzles”, being good at solving puzzles will become a less valuable skill.
This doesn’t mean there will be less jobs for engineers (I actually think AI means we’ll have more developers), I think it just means the skillset and how the job looks day-to-day will evolve.
If I was an engineer (I am definitely not), these are the skills I’d be trying to hone, refine and leverage in the next era.
Having strong empathy for customer problems
Being able to clearly articulate those problems
Designing robust systems and software architecture
Context engineering, orchestrating AI and working with agents/LLMs
Those skills feel very AI-proof. At least for now.
