Your AI Strategy Isn’t Stalled. You Just Skipped the 90-Minute Conversation Everyone Avoids.

Most companies I speak with say the same sentence: ‘We’re behind on AI.’ Very few are behind on AI. They’re behind on deciding.

The pattern is everywhere: a Slack channel called #ai-strategy, a Notion doc with dozens of use cases, maybe a paid consultant on retainer — and zero workflows in production that touch revenue.

The pattern is identical every time

Smart founder reads a McKinsey report. Founder forwards report to leadership. Leadership runs a workshop. Workshop produces a roadmap. Roadmap gets parked because nobody on the roadmap is also accountable for shipping it. Six months pass. The founder reads another McKinsey report.

This is not an AI problem. This is the same problem CRMs had in 2009, mobile had in 2013, and ‘digital transformation’ had in 2019. The technology is downstream of a missing decision.

The 90-minute fix

Block 90 minutes. Three people in the room — no more, no less:

• The founder or CEO (the person who can override budget).
• The person who actually owns the P&L of the workflow you want to change.
• One builder — engineer, ops lead, or external implementer — who can ship something in 30 days.

On a single page, write four things. Out loud. With names attached:

1. The one workflow we are changing first. Not three. One. Pick the one closest to revenue.
2. The human who owns the ‘yes’ — the one person who can say ship it without a second meeting.
3. The kill-switch — the metric or moment that means we turn it off. (No kill-switch = no launch.)
4. The 30-day deliverable. Not a deck. A working slice — even ugly — that a real customer touches.

That’s the entire exercise. If you can’t fill out the page in 90 minutes, the bottleneck isn’t the technology. It’s that nobody in the room has the authority to choose. That is a different, harder, much more useful conversation to have.

Why this works when frameworks don’t

Strategy decks are decoupled from execution by design — that’s what makes them feel safe. A named owner with a kill-switch is the opposite of safe. It’s a forcing function. Teams that adopt this exercise consistently ship working AI pilots in under 30 days. The deck would have been 80 pages and shipped nothing.

“The companies winning at AI right now aren’t the ones with the best models. They’re the ones with the fewest meetings between a good idea and a working prototype.”

What to do tomorrow morning

Cancel the next AI strategy session. Send this post to the two other people you’d put in the room. Book 90 minutes. Fill out the page. If you can’t — and many can’t, that’s the actual finding — then you know exactly what to fix first, and it isn’t the AI.