With AI, software teams everywhere can now produce significantly more volume. We've all seen it: the amount of code being written has jumped, almost everywhere at once.
But does volume translate to value?
Look at the numbers most organizations report about AI: adoption rates, agents running, percent of code written by AI, pull requests merged. Every one of them is climbing, and every one of them measures the Tool. They tell you the Tool is in place and being used. They can't tell you whether anything got better.
In the House of Software Value Delivery, you change Tools to move Attributes. The Attributes are indications of the value of the software. Measure the Attributes, not the Tools.
So ask which Attribute the AI was supposed to move. Often the Attribute leaders want is Timely, and that one usually does move. Features get delivered faster. But that needs to be paired with other Attributes, like Maintainable, Intuitive, and Desirable.
Are changes getting easier and safer to make? More code delivered means more code to maintain, so Maintainable may be moving in the wrong direction.
Are customers able to understand and use all of the functionality changes being pushed out to them? Do they even want those features? Code and feature volume can double without the software getting any more Desirable, and it can get less Intuitive along the way. I've heard from leaders whose customers don't want a flood of new features. For those customers, doubling the volume means shipping more of what nobody asked for.
Measure adoption to confirm the Change happened. Then measure the Attributes to find out whether it was worth it. Volume is a Tool number. Value shows up in the Attributes.
