Here’s an uncomfortable pattern I’ve seen across two decades of technology investment, sharpened by everything AI has taught us recently.
Most large software projects are described as technology projects. They mostly aren’t. Strip away the platform and look at where the effort and value actually sit: the business is forced to examine its processes and finally agree how work should be done. Forced to agree the rules for its data, and clean it up for migration. Forced to name who owns which process and who decides business outcomes. That’s the transformation. The software is the occasion for it, and often an expensive occasion.
Which raises an awkward question I now put to every investment case. If the clarity is the value, why does it need the project? An operational leader who mapped their processes, named owners, agreed data rules and cleaned the critical data, without any new platform, would bank a startling share of the improvement at a fraction of the cost. We rarely do it because a project deadline supplies the forcing function that ordinary management somehow doesn’t. We’re effectively paying vendors millions to make us do our own homework.
AI raises the stakes on this old observation in a specific way. That homework, understood processes with owners, agreed data rules, clean data, clear outcome accountability, is exactly the foundation agents need. Do it and you’ll frequently discover AI can deliver material benefits on top of it directly, without the platform program you assumed you needed.
So before we invest now, I want answers. Who owns the processes this touches, and can they describe them? What are the data rules, and who arbitrates them? What outcome improves, by how much, owned by whom? And the killer: how much of the benefit would we capture by focusing on people, process, and data, without spending money on technology?
If those answers don’t exist, the project isn’t ready, and its business case is partly a fiction. If they do exist, you may need less technology than you think.