Execution is everything… or nothing?

March 27, 2025

There is an old business adage: execution is everything. Ideas are a dime a dozen and do grow on trees, but making them happen and turning them into a viable business is a different story altogether. But in the age of artificial intelligence, does this maxim still hold up? Or has execution suddenly become nothing?

The legendary Intel chief executive Andy Grove said it. Amazon founder Jeff Bezos said it. Many others have too: execution is everything1. Often the phrase is introduced together with the claim that ideas are nothing. Execution refers to the effective implementation of strategies and objectives. It is the how to the idea’s what. A million-euro idea remains worthless until made real. What bridges the gap between an idea and the potential outcome is execution, a messy weave of people, roles, structures, conference rooms, buildings and customers, commonly organized into functions like finance, sales, marketing, compliance and support. And development of course. Any technology powerful enough to lighten this overhead will, in time, be adopted.

A striking feature of modern artificial intelligence, especially large language models, is their ability to turn words into working software almost like magic. This points to a comming wave of programming commoditisation. Once a scare and costly branch of cognitive work is rapidly becoming infinitely accessible at ever lower marginal cost. With the rise of vibe coding almost any digital product can seemingly be cloned on the spot without much effort24. The barrier to entry has vanished. A straightforward enough concept can now be replicated instantly by anyone anywhere anytime. It's sounds like a speed run through twentieth centrury manucaturing, littered with cheap mass-produced knock-off products from dubious companies in far-away lands. Some companies not only endured, but thrieved in this environment, as did the non-manufacturing functions supporting their survival. The allure of premium products, with uniqueness sometimes only conjured up by crafty sales and marketing deparments, turned originality and scale up and outsourcing of manufacturing into winning concepts and the mentioned sales and marketing functions ubiqutous.

To stay ahead of imitators using AI, programmers are embracing the same tools and dialling development velocity up another notch. Just three decades ago developers relied on printed books and boxed documentation, the web and searching moved most answers to the finger tips of the programmers and raised speed another notch, and community stores of knowledge like Stack Exchange together with systems such as Jira and git amplified the effect further. And soon most, if not all, parts of software development will include a component of AI during generation, enhancement, refactoring, bug fixing or testing.

Will ninety percent of software really be produced by AI during 2025, as Anthropic chief executive Dario Amodei predicts3? His timeline may prove accurate, though perhaps not for the reasons first assumed. The programmer’s role is shifting from executee to executor, from coding specialist to domain specific generalist, from hands-on producer to conductor of coding workflows. Programming may also simply be the first cognitive field to undergo this form of commoditisation, thanks to a vast public archive of source code. Once firms recognise that other branches of knowledge work share similar patterns and begin to harvest equivalent data, the same transformation will follow elsewhere. Tasks we now regard as vital parts of execution will be automated one by one. One day execution itself may truly be nothing, while orchestrating that execution is everything.

Andreas Påhlsson-Notini

andreas@nial.se