Project Failure Files: AI Amplifying Chaos

In Episode 84 of the Project Failure Files weekly webcast, Sharon and I kicked off a 5-part miniseries exploring leadership and management issues with the rapid rise of AI technology. We’ve been talking around the topic throughout our episodes, but while both of us have spent much of the last 2 years leading AI strategy and training workshops, we felt it was needed to address the AI-specific topics on their own. So, for this first episode in the miniseries, on our focus was “Automating a Mess.” We define this as the expensive leadership mistake of layering AI on top of undocumented, inconsistent, fragile workflows—and expecting intelligence to compensate for a lack of clarity. The discussion frames this as a familiar pattern (like “just add search” or “just add collaboration”) where technology gets blamed for problems that were actually caused by weak foundations. AI doesn’t fix broken systems; it accelerates whatever is already happening—good or bad.

The episode explores how speed can create a false sense of progress while quality quietly degrades. In messy environments, AI surfaces outdated, duplicated, or conflicting content faster than teams can validate it, and overconfidence in outputs can erode trust. The cultural impact shows up as failed pilots, misaligned expectations, and “AI didn’t work” narratives that poison future innovation—sometimes alongside security and compliance surprises when AI makes previously hidden content discoverable.

The path forward is deliberately unglamorous: stabilize before you optimize. That means documenting workflows, clarifying roles and decision rights, improving data hygiene, and building governance and measurement into the rollout. A simple readiness test is whether a brand-new employee could run the process using what’s documented—if not, automation will just fill gaps with guesses. The call to action is to pick one high-interest process, run a ruthless stability check, fix the cracks, then automate with oversight and measurable outcomes.

Enjoy the episode!

Be sure to tune in next Monday, April 6th at 9am Pacific for the second in our 5-part miniseries, where we’ll discuss how without clear AI guidelines and guardrails, organizations risk data exposure, compliance failures, and reputational damage. Hope you can join us on our NEW YouTube channel (please subscribe!) or find us on LinkedIn.