Your company doesn't need an AI strategy

Sometime in the last two years, “What’s our AI strategy?” became the most popular question in boardrooms. It sounds serious. Forward-thinking. Like the kind of question that justifies a six-figure consulting engagement and a 90-slide deck.

It’s also the wrong question.

Here’s why: an “AI strategy” frames the technology as the starting point. You begin with AI and then go looking for places to apply it. This is exactly backwards. It’s like asking “What’s our electricity strategy?” in 1920. The answer isn’t about electricity. It’s about what you’re trying to do and whether electrification helps you do it.

Most companies asking about AI strategy don’t have a technology problem. They have a clarity problem. They can’t clearly articulate what their actual bottleneck is. They can’t point to the specific process, decision, or workflow that, if improved, would meaningfully change their business. And without that clarity, AI becomes a solution looking for a problem.

The companies getting real value from AI didn’t start with “How do we use AI?” They started with “What’s actually broken?” and AI happened to be the fix.

I’ve seen companies spend months evaluating AI vendors before spending a single hour mapping out where their people waste time. They compare models and platforms and pricing tiers without first asking: what is the actual work that needs to happen, and where does it break down?

The pattern is always the same. A company buys an AI tool. It gets deployed to a team. The team uses it for a few weeks. Usage drops. Someone asks why. The answer is usually some version of “it didn’t really fit how we work.” Not because the tool was bad, but because nobody figured out the workflow first.

This is the clarity problem. And it’s not new. It predates AI by decades. Companies have been buying technology to solve problems they haven’t defined since the invention of enterprise software. CRM systems bought by companies that don’t understand their sales process. Analytics platforms bought by companies that don’t know which metrics matter. AI is just the latest in a long line of expensive answers to unasked questions.

The fix isn’t complicated, but it requires a kind of honesty most organizations resist. You have to sit down and ask: where are we slow, and why? Where do decisions get stuck? What do our smartest people spend time on that doesn’t require their intelligence? Where does information get lost between teams?

These are boring questions. They don’t make for good conference talks. Nobody gets promoted for saying “we spent three months mapping our internal processes before buying anything.” But the companies that do this, the ones that start with the problem instead of the solution, are the ones that actually get value from AI.

Because once you’ve identified the real bottleneck, the technology choice often becomes obvious. Sometimes it’s AI. Sometimes it’s a simple database query that nobody thought to write. Sometimes it’s firing a tool and hiring a person. Sometimes it’s removing a process rather than automating it. You can’t know until you’ve done the diagnostic work.

There’s another version of this problem that’s more subtle. Some companies do know their bottleneck, but they describe it in terms that are too abstract to act on. “We need to be more efficient.” “We need better decision-making.” “We need to innovate faster.” These aren’t problems. They’re wishes. They’re too vague for any technology to address, AI included.

A useful problem statement is specific and observable. “It takes our underwriting team 4 hours to process an application because they manually check 12 different data sources.” That’s something you can fix. “We need to use AI to transform our underwriting process.” That’s something you can make a deck about.

The irony of the AI hype cycle is that the technology itself is genuinely powerful. Large language models, code generation, automated analysis: these tools can create real value in the right context. But “the right context” requires knowing what you’re trying to accomplish with enough precision that you can tell whether the tool is working.

So the next time someone asks “What’s our AI strategy?”, redirect. What’s our bottleneck? Where do we lose time, money, or quality? What would change our business if it were 10x faster or 10x cheaper?

Start there. The strategy will follow. Or it won’t need to.