Are We Optimizing for Speed, or for Value?

There’s a trend in tech right now to chase developer productivity. With AI coding tools getting better so fast, it feels like the obvious thing to optimize.

But I think we’re missing something bigger.

It’s not just about building faster. It’s about helping engineers know when not to build something at all. When not to start a new system from scratch. How to keep things simple. How to think clearly about the problem before getting excited about the solution.

This was already a problem before AI. Teams would spend months building systems that, within a year or two, were deprecated or questioned for their actual value. At that point, it’s fair to ask: was that truly a success?

Companies are not looking for more systems and more code. They are looking for outcomes. They care about business value. Dollars saved, revenue generated, time reduced, and happier customers.

If teams repeatedly find themselves working on deprecations, consolidations, or integrations, it is worth examining where the productivity is really being lost.

I do think engineers should have a product mindset. But I also think it is very easy for people who love technology to over-index on the fancy solution rather than choosing the right problem to solve, or deciding that a problem is not worth solving at all.

AI may not solve this. In fact, it might amplify it. AI gives us a faster racecar. But a fast racecar with a less than competent driver is not a recipe for success but rather disaster. When tools are very agreeable and make it easier to build, we may end up building more things that never needed to exist. More duplication. More complexity. More systems with limited business value.

I am optimistic about where AI will take software engineering. But I am equally cautious about the risk of creating an abundance of low-value work.

Maybe productivity as just “more output faster” isn’t the right metric at all. We should think more about efficiency in the real sense: reducing waste and focusing on what matters, not just doing things faster.

For me, the main reason AI makes me more productive is not just speed, but because it amplifies my judgment. Context, experience, and clarity of thought still matter most. The tool does not magically make you productive. What makes you productive is how you use it and the judgment behind it.

Are we optimizing for speed, or for value?