Your Words Become Your AI Policy: Crystal Pee Bowman on Rules a Small Team Will Use

When you hear "AI policy," you probably picture a list of don'ts. Don't paste in customer data. Don't use tools we haven't approved. Don't get us in trouble.

Dr. Crystal Pee Bowman would call that half a policy.

Crystal is External Affairs Manager and workforce development lead at Integer Technologies, a defense research company. She has a PhD in engineering education from Virginia Tech, where she studied how early-career engineers adapt when their workplace changes. I heard her on a workforce panel at the NH Tech Alliance Innovation Summit in Nashua this week, and we've since connected on LinkedIn.

The panel was about AI governance, which is a fancy name for the rules a business sets for using AI. It sounds like a big-company topic. Crystal's ideas fit a shop of six people just as well.

Tell your team how to use it

Crystal pointed to Gallup research showing that companies with an actual AI use policy see more productivity and more trust.

Gallup's numbers back her up. When employees strongly agree their leaders have shared a clear plan for AI, they're 4.7 times as likely to feel comfortable using it in their role (Gallup). Gallup's 2026 look at AI and productivity found the same pattern: a clear strategy and responsible-use guidelines go hand in hand with people using AI more and getting more out of it.

"What compliance does just tells you what not to do. It doesn't tell you how to do it and how to move it forward."

A compliance list says don't paste customer data into ChatGPT. A use policy says here's what we use AI for, here's how we do it well, and here's who to ask. At Integer, she said, people are expected to use AI to help with their tasks, properly and in context. The policy is there to make that possible.

For a small business, one page is plenty:

1. What we use AI for. Three or four jobs, written plainly.
2. What tools we use. The ones you've chosen, so nobody's guessing.
3. What never goes in. Card numbers, passwords, private customer details. I keep a fuller list in what you should not use AI for (yet).
4. Who to ask. One name.

Your words are already a policy

This was the line I wrote down twice. "Language becomes practice long before it becomes policy."

Crystal's example was phrases like "AI headcount reduction." Say something like that often enough and it seeps into how decisions get made, whether anyone wrote it down or not.

In a small business, the owner's words carry even more weight. Talk about AI as a way to cut people, and your team will hear that and act on it. Talk about it as a way to take the busywork off their plate, and they'll hear that too.

Progress with a purpose

Crystal sees AI as a tool, and she wants a reason before anyone picks it up. "I still think the principles of business still apply. What is the business case?"

Her phrase for it is "progress with a purpose." People, she said, want to know where AI is useful in their own work. Blanket AI use leaves them cold. She compared it to how businesses figured out Excel.

So name the job first. Quoting, scheduling, the weekly report. Then ask whether AI makes that job easier.

Two truths at once

When the moderator asked how to handle employees who resist AI, Crystal started with honesty. "Two truths are true." AI does make people more productive. And it isn't automatically good for everyone.

Her advice was to drop the sweeping statements about what AI will replace. Talk about how roles will be redefined, and be clear about what you're trying to do. People can tell when they're being sold.

Is it speed of sound in water, or in air?

Crystal pushed for AI literacy, meaning people understand their work well enough to catch AI's mistakes. Knowing which buttons to press is the easy part.

Her example came from Integer's acoustics research. If AI calculates something with the speed of sound, did it use the speed in water or in air? "You got to understand when something is wrong. So the fundamentals still need to be there."

Your version might be a quote with the wrong tax rate, or a reply that promises a delivery date you can't hit. The person using AI needs to know the work well enough to catch it.

Keep a bench

Crystal also raised something most AI conversations skip. If AI does all the beginner work, where do beginners learn? She cited a Federal Reserve figure showing unemployment for young workers running above the national rate, a reversal of the usual pattern.

She put it in basketball terms. The Golden State Warriors won in 2015, she said, because they had a deep bench. Businesses need one too.

Sam Pause, Chief Innovation and Brand Officer at Mascoma Bank, made the same point from the next seat. Sam had watched the bank's CTO use AI to write code during a meeting (the CTO called it "a junior programmer") and wondered where senior programmers will come from. Sam's own know-how came from 35 years of doing the work.

In a small business, your bench might be one new hire. If AI handles the work they'd have learned on, give them other ways to learn it. Let them check AI's output against your own. Bring them into the decisions.

Where the panel landed

Brian Bouchard, an employment attorney at Sheehan Phinney, moderated. Sam argued that good rules help people experiment, because people get curious once they feel safe. Vince Berk, a tech investor at Apprentis Ventures, took the other side: teach people the basics of handling data, then trust them, since rules always lag behind the tech.

They all came back to trust.

Near the end, Crystal reached back to management thinker Mary Parker Follett, an idea she'd picked up from a book by future-of-work researcher Kelly Monahan: let the situation decide what you do. Figure out the problem, then look at what you have to solve it.

"Good business is still good business."

Put that at the top of your one-page policy and you're off to a good start.

Business Brain gives your AI the context to follow that policy: your voice, your customers, and how you do things. If you'd rather write the policy together, get in touch. I enjoy working on these.

Also from the summit: How to Turn AI Skeptics Into Advocates, with Melanie McDonough.