Brian Beck runs a cybersecurity and AI consulting firm out of Indianapolis, and during our conversation in The Next Build podcast, he shared something with me that stuck with me for weeks. Clients call him constantly, multiple times a week, and they all open the conversation the exact same way. “I want AI”. And his response is always identical too. Okay, to do what?
Most of the time, there’s no real answer. Just “something”. That single word is doing a lot of damage in small businesses right now. Leaders are excited, budgets are getting approved but nobody stops to ask the obvious question. If you’re thinking about AI adoption for your small business, that question is exactly where you need to start, and skipping it is the single most common mistake Beck sees in his work.
Why "Something" Isn't a Strategy
Here’s the thing. Excitement is not a plan. You can feel genuinely thrilled about a new tool, get a rush from watching it do something clever, and still have zero idea whether it’s actually helping your business. Brian put it bluntly that adrenaline hit is great, but from a business point of view, how do you monetize a feeling?
A lot of companies give AI licenses to their whole staff because everyone was asking for it. No goals attached. No timeline. No way to measure whether it worked. Brian knew of one organization that handed out hundreds of copilot licenses for exactly this reason, and when he asked why, the answer was basically “people wanted it”. That’s not a decision. That’s peer pressure with a budget behind it.
Before you spend a dollar on AI adoption, figure out what you actually want changed. Faster customer response times? Fewer hours spent on repetitive writing tasks? Or a tighter process? Pick something specific that you’d know after three months, whether it actually happened.
The Foundation Nobody Wants to Check
This part gets skipped constantly, and it’s the part that actually matters most. Brian describes himself, half-jokingly, as “the concrete guy”. You wouldn't build a building without checking the foundation first, and cybersecurity is the foundation everything else in your AI adoption strategy sits on top of.
If your systems have cracks in them already, adding AI on top doesn’t fix that. It just gives those cracks more speed and more access. A tool that’s compelled to always produce an answer, even when it doesn’t know it, will happily go pull information from the open web if your internal data isn’t well organized. Sometimes that means hallucinated answers pulled from questionable sources. Sometimes it means information leaking somewhere it shouldn’t.
So first of all, check. When was your last real security audit? Who actually manages it, one overworked person or an actual process? If you’re not confident in the answer, that’s your starting point, not a reason to roll out AI.
Picking a Platform Without Falling for the Hype
Once the goal and the foundation are settled, then you can talk tools, and this is where Brian’s advice becomes genuinely useful. He’s not a hype guy. He waits for version two or three of anything before touching it, foldable phones included, and he brings that same patience to AI platforms.
His default recommendation for most small businesses already running Microsoft 365 qis Copilot, mainly because it inherits your existing security permissions instead of creating a , and brand new access point to manage. If you are in a Google environment, Gemini plays a similar role. Neither pick is about brand loyalty. It’s about reducing the number of separate systems your team has to secure and babysit.
One detail worth knowing: Copilot’s adoption jumped to roughly 52% in a single year, and the spike happened right when Microsoft made Claude available on the back end. People weren’t loyal to the wrapper. They wanted access to specific models they’d already fallen for elsewhere, and Copilot finally gave them that without forcing a switch. That’s a pretty good sign that the smartest move isn’t picking a favorite AI brand. It’s picking the setup that keeps your access and your data in one place instead of scattered across five logins nobody fully tracks.
Getting the Whole Team On Board
According to Brian, adoption fails when it’s left to trickle down from just one or two enthusiastic employees instead of coming from leadership with an actual plan. He’s seen teams burn through an entire AI budget because two people were using every credit while the rest of the staff sat on the sidelines, half-convinced the tool was coming for their job.
That fear is real and pretending it isn’t doesn't help anyone. Leadership has to say, out loud, why this is happening and what it’s meant to complement rather than replace. If you skip that step, you’ll get a company split between people who are terrified and people who are using AL a little recklessly with no clear direction.
Do This Before You Adopt Anything
If you are serious about AI adoption for your business, don’t start with a tool. Start with a plan. One clear goal, an honest look at your cybersecurity foundation and a platform choice that reduces complexity instead of adding to it. Slow down enough to get those three things right, and you’ll move past the business still floundering around with “something” as their entire strategy.