A signal, not a headline

When a company reports that AI is writing more than 65% of its new code — and trims headcount in the same announcement — it's tempting to read it as a layoff story. It's more useful to read it as a signal about where engineering is heading, and how fast.

The bottleneck has moved. For years the constraint was "can we build it?" The constraint now is "can we evaluate what was built — accurately and quickly?" That's a different muscle, and most teams haven't deliberately trained it.

The job is changing shape

The day-to-day of a strong engineer is shifting in a specific direction:

Less time writing routine code. Boilerplate, glue, the obvious CRUD layer — increasingly generated.
More time reviewing and directing. Catching the subtle thing the model got confidently wrong.
System design and critical thinking as the core output. The parts that require holding the whole system in your head are exactly the parts AI handles worst.

None of this makes engineers less valuable. It raises the bar on what "valuable" means. The most important engineers in the next few years won't be the fastest typists. They'll be the sharpest reviewers — people who understand a system deeply enough to know, quickly, when the generated answer is subtly off.

Reviewing is harder than writing

There's an uncomfortable truth here: evaluating code well is often harder than writing it. Writing forces you to reason through every line. Reviewing tempts you to skim, especially when the output looks plausible — and AI output almost always looks plausible. Building review as a real discipline means slowing down at exactly the moments the tooling encourages you to speed up.

Stop measuring the wrong thing

If your team still measures developer performance by lines of code or raw output volume, you're optimising for the thing AI now does for free. Worse, you're rewarding people for generating more surface area to review.

Better signals in a generation-heavy world:

Defect rate and escaped bugs — is the reviewed output actually holding up in production?
Time-to-first-meaningful-review — how fast can the team turn generated code into trusted code?
Rework rate — how much shipped code comes back?
Quality of system design decisions — harder to quantify, but the highest-leverage thing a senior engineer contributes.

The practical move is to start building the review muscle now, deliberately. Make architectural review a standing part of every cycle. Invest in static analysis and strong test coverage so reviewers spend their attention on judgment, not mechanics. Pair junior engineers with seniors specifically on evaluation, not just authoring.

The takeaway

The teams that treat reviewing, directing, and system design as the core engineering skill — and instrument their teams around it — will outship the ones still optimising for typing speed. This is a skills shift worth investing in before your competitors finish doing it.

If you're rethinking how your engineering team operates in this new shape, we can help. We're here to help founders and teams design and build digital products that scale with you, not slow you down. If you're looking to build something, get in contact with us today.