Why most companies are getting AI wrong

Most of the AI conversations I have with leadership teams sit on one of two extremes.

On one side: "We need an AI strategy by Q3." Big committee. Big budget. Nothing ships.

On the other: "Our team already uses ChatGPT, we're fine." No real workflow change. No measurable lift.

The middle is where the actual money is. It's also, frankly, boring — which is why it's the path that gets skipped.

The boring path

Pick a workflow that already exists. Watch someone do it. Find the bit that's repetitive, slow, or annoying. Build a small AI-shaped thing that removes just that bit.

That's it. That's the whole playbook for the first six months of any company's serious AI work.

The reason it works isn't because the AI is clever. It's because the change is small enough that:

• People actually adopt it
• You can measure whether it helped
• You can rip it out if it didn't
• You learn something real before betting bigger

What "AI strategy" usually means

When a leadership team asks for an AI strategy, they usually mean "help me feel less anxious about AI." That's a legitimate need — but it's a therapy need, not a consulting one.

What I try to do instead is replace the abstract anxiety with three or four concrete bets, each small enough to ship in a quarter.

That's it for this post. More to come.