Code Is Almost Solved. Product Is Not.
The bottleneck has shifted.
For decades, software companies were constrained by engineering capacity. You had more ideas than people who could build them. The limiting factor was implementation — writing code, shipping features, keeping systems alive.
That constraint is dissolving. AI can now reliably generate code, implement features, and refactor systems. Low-level code generation is, for most practical purposes, solved.
But something else has become visible in its absence.
What’s actually hard
Most failures don’t come from bad code. They come from:
- Unclear product intent
- Missing architectural constraints
- Weak specifications that drift over time
- Systems that work once but break on iteration
When AI amplifies your development capacity by 10x, it also amplifies the cost of getting the product definition wrong by 10x. A strong system produces faster progress. A weak system produces faster entropy.
But it goes deeper than just “write a better spec.” Production-grade systems thinking and DevOps are not solved. Defining product, user experience, and user interface is still an open field. And the hardest part — getting autonomous AI to diligently abide by constraints, without human control, review, or oversight at every step — is an ongoing, unsolved challenge.
The real work
The role of the engineer is shifting. Not disappearing — shifting. The valuable work is no longer writing code. It’s:
- Defining product intent clearly enough that AI can execute it
- Encoding constraints (architectural, product, operational) that prevent drift
- Maintaining coherence as systems evolve, not one-shot after one-shot
- Keeping products stable over duration, meeting user expectations consistently
This is the work I’m focused on now. Not faster prototyping, but reliable translation of product intent into working, durable systems.
The spec is not documentation. The spec is the product. And the constraints are not optional — they’re the difference between a system that works and one that slowly falls apart.