Here's why more AI training won't fix it — what the rollout won't admit

You got the login. You sat through the training. And your week looks exactly the same as it did the month before. That's not because you failed the course — it's because the course was never going to touch the thing that's actually slow.

Companies keep treating AI like a skills problem you can solve with a training session. But you learn the tool and get sent right back into the same meetings, approvals, and sign-off chains that made the work slow in the first place. This memo is about why the training "works" while nothing changes — what handing out the logins actually bought, what the course really closed, and the one question that tells you whether your rollout is real or theatre.

ℹ️ AI voice disclosure: This episode uses an AI-generated voice (ElevenLabs). The host is a pseudonymous byline persona (Reid Conlin).

In this memo

1. 0:00 Cold open
2. 1:27 The login that looked like the rollout
3. 2:51 The course that was supposed to fix it
4. 4:05 The same slow machine
5. 5:38 What actually moves the work
6. 7:29 The whole memo, backwards

Sources

• EY, 2025 Work Reimagined Survey (88% use AI, 5% transform their work; the training dose-response):
• OpenAI, The State of Enterprise AI 2025 (the 6x–17x usage gap; unused features):
• Deloitte, State of AI in the Enterprise (leaders name "skills" as the #1 barrier, then run training):
• McKinsey, Superagency in the Workplace (employees want more training; leaders underestimate adoption):
• CIO, "You can't train your way out of the AI skills gap":
• Moderna × OpenAI case study (100% adoption as a change project, 750+ custom GPTs):
• Klarna customer-service reversal, Bloomberg (CEO: cost became "too predominant a factor"):

Watch on YouTube ↗ Subscribe ↗