If I had a dollar for every CEO who told me their board is asking about AI, I'd have a lot of dollars. And if I had another dollar for every time that conversation led somewhere concrete and useful — I'd have considerably fewer.

The problem isn't that mid-size companies aren't interested in AI. They're intensely interested. The problem is that the conversation is happening at the wrong altitude. Boards are asking about AI as a strategic imperative. Leadership teams are scrambling to have an answer. And somewhere in the middle, people are signing up for tools they don't fully understand, running pilots that don't connect to business outcomes, and building anxiety about falling behind competitors who may or may not actually be further ahead.

Let me offer a more grounded path forward.

Start with honest assessment, not ambition

The first question in any serious AI conversation isn't "What can AI do for us?" It's "What are we actually capable of today?" AI doesn't create capability from nothing — it amplifies what you already have. If your data is a mess, AI will give you faster, more confident wrong answers. If your processes are poorly defined, AI will automate chaos at scale.

An AI readiness assessment covers three things:

  • Data: What data do you have, where does it live, how clean is it, and who controls it? Most mid-size companies discover in this process that their data is more fragmented and less reliable than they assumed.
  • Processes: Which workflows are well-defined enough to benefit from automation or AI assistance? Vague processes don't improve with AI — they just fail faster.
  • People: What's your team's current relationship with technology? Adoption is as much a culture problem as a technology problem.

This assessment takes weeks, not months — and it's the most valuable investment you can make before spending a dollar on AI tools.

Find the real opportunities — not the obvious ones

When I ask leadership teams where they want to use AI, the first answers are almost always the same: customer service chatbots, marketing content, maybe some kind of analytics dashboard. These aren't bad ideas, but they're also not usually where the real value is.

The highest-value AI applications in mid-size companies tend to be less glamorous:

  • Internal knowledge retrieval (finding answers in your own documents and systems)
  • Operational reporting and anomaly detection
  • First-pass review of contracts, proposals, or compliance documents
  • Automating data movement between systems that don't talk to each other
  • Supporting staff in repetitive, high-volume tasks — drafting, summarizing, classifying

The best AI use cases are usually the ones where someone on your team says "I spend two hours a day doing this manually, and I hate it." Start there.

The goal is to find applications where the effort is well-defined, the benefit is measurable, and the risk of a wrong answer is manageable. That's a very different profile from "let's build a customer-facing AI."

Data governance can't be an afterthought

This is the part of the AI conversation that most vendors skip, because it's not exciting and it doesn't help them close a sale. But it's one of the most important things to get right before you start.

When you use AI tools — especially cloud-based ones — data moves. It goes to servers you don't control, processed by models you may not fully understand, potentially used to train systems in ways you haven't agreed to. For most productivity use cases, this is manageable. For data involving your clients, your financials, or anything regulated, it requires careful thought.

At minimum, before deploying AI tools broadly in your organization, you need:

  • A clear policy on what data can and cannot be used with AI tools
  • An understanding of the data handling practices of any vendor you're using
  • A communication plan for employees so they know what's allowed
  • Some mechanism to see what's actually being used (you'd be surprised)

This isn't about being overly cautious. It's about being deliberate. The organizations that will do best with AI are the ones that move thoughtfully, not the ones that move fastest and deal with the consequences later.

The pilot trap — and how to avoid it

Many mid-size companies have run AI pilots. Far fewer have scaled them. The pattern is familiar: a motivated team runs a proof of concept, gets impressive demo results, and then… nothing happens. The pilot sits. The team moves on. The vendor keeps calling.

The pilot trap happens for a few reasons:

The pilot solves the wrong problem

If the pilot isn't addressing something the business genuinely cares about measuring and improving, there's no organizational energy to push it forward. "This is cool" isn't a business case.

Success isn't defined upfront

What does it mean for the pilot to succeed? If you can't answer that clearly before you start, you'll never be able to make the case to scale it.

No one owns it

AI pilots need a business owner — someone whose job is on the line for whether it works. Without that, pilots become research projects. Research projects don't become business capabilities.

The fix: Before launching any AI pilot, define three things: the business problem it's solving, the metric you'll use to evaluate success, and the person accountable for the outcome. If you can't define all three, don't start the pilot.

What a real AI strategy looks like

A practical AI strategy for a mid-size company isn't a 50-page document. It's a set of clear decisions:

  • Where we'll focus first — based on readiness and business value, not hype
  • What we won't do yet — because scope control is as important as ambition
  • How we'll govern data — what can and can't be used with AI tools
  • How we'll measure progress — concrete metrics, not vibes
  • How we'll build capability — skills, tools, and processes that compound over time

It's a living document, not a one-time exercise. The AI landscape moves fast. Your strategy needs to be able to move with it.

The right pace is deliberate, not slow

I want to be clear: I'm not arguing for caution at the expense of progress. The organizations that do nothing with AI will fall behind. The question is what "doing something" looks like.

Deliberate means: knowing why you're doing what you're doing, having the foundations in place to do it well, and being able to measure whether it's working. That's not slow. That's how you actually build something that lasts.

If you'd like help thinking through what an AI strategy looks like for your organization specifically — based on where you actually are, not where you want to be — book a free 30-minute conversation. That's where it starts.