Writing · The AI shift
The three things real AI transformation needs
The ROI headlines around AI rollouts have turned: lately it's more disenchantment than hype. I don't think that's because the technology is a bust. It's because rolling out AI is sneakily difficult.
Generative AI is emergent and non-deterministic, so capturing real benefit inside an organization takes three things at once.
Fluency in generative AI, and (almost more important) deep, data-driven operational-intelligence experience. You have to know what deterministic guardrails an inherently non-deterministic technology needs.
Someone who's good at digging to find the problems that actually matter and can be meaningfully helped by this tech. There's no playbook; you invent the workflow, build the cheap version, and watch it run.
The best solution delivered badly still fails. Adoption is earned, not announced.
Miss one and the failure is predictable:
- Miss the technical skill, and you're paying the stupid tax at scale.
- Miss the inventiveness, and you keep up, at best; you don't lead the industry.
- Miss the change leadership, and it dies; it never gets adopted.
Most people bring one of these. Some bring two. The rare combination is all three at once, and that's what "AI transformation" quietly requires.