Project Failure Files: Ignoring the Business Role in AI
In Episode 86 of the Project Failure Files weekly webcast, Sharon and I argue that AI adoption fails (or underdelivers) when it’s treated like a standard IT rollout—licenses assigned, training delivered, “deployed” checked off, and everyone moves on. The core point is that AI isn’t just another tool in the stack; it changes how decisions get made, how workflows operate, and how teams collaborate—so the “old playbook” produces incremental results when the opportunity is transformational.
We highlight the organizational failure modes that follow: business teams waiting on IT, siloed pilots that look great on slides but don’t scale, and an adoption plateau where people try it once and then don’t know how it fits into daily work. There’s also a strong warning that locking AI down too tightly doesn’t stop usage—it simply drives shadow AI, sending experimentation (and often corporate data) outside of governed environments.
The solution theme is shared enterprise ownership: IT leads governance, security, compliance, and architecture, while business leaders own relevance, accountability, and measurable outcomes. That requires defining the problems AI is meant to solve, agreeing on success metrics, enabling safe experimentation, and training leaders—not just end users—because leaders decide what the organization does with AI’s output.
Enjoy the episode!
Be sure to tune in next Monday, April 20th at 9am Pacific for the fourth in our 5-part miniseries, where we’ll discuss how executives are under pressure to show fast returns on AI investments, but meaningful productivity improvements require learning curves, experimentation, and behavioral change. Hope you can join us on our NEW YouTube channel (please subscribe!) or find us on LinkedIn.
