This LinkedIn post from Isaac C. is about software development and an individual developer, but it applies just as well to ML/AI and a company. And it's precisely why I've long encouraged companies to develop a plan before rushing into their AI adventure.

When I say "AI strategy" (or way back when, "data strategy") some people assume that I mean some kind of mission statement. Not at all. This should be a detailed plan that includes:

• "What are we doing?"
• "Why are we doing it? And why this way?"
• "Why should we not do this?"
• "What does success look like?"
• "Under what circumstances do we stop?"
• "Why do we expect this approach will bear fruit?"
• "What are some ways this can go wrong?"
• "Who's doing what?"
• and most importantly – to borrow Isaac's post – "no assumptions."

This approach works for small projects as well as large, company-wide matters. In fact, it's especially important for the small projects because those can quietly drift for a long time without anyone noticing.

Having a plan means knowing where you're trying to go and how you'll get there. Which means you'll also know when you're not getting there.