If you’re thinking about bringing someone in to “handle AI” for your business, read this first. The hire may make sense eventually. But there’s a question that usually needs an answer before it does.
A business owner we spoke with recently was planning to hire an AI specialist for about $200,000 a year. He kept hearing about what AI could do, other owners were experimenting with it, and he didn’t want to fall behind. His reasoning was simple: if AI is the future, he needed someone who understood it.
Before he made the hire, our co-founder, Nesanel Moeller, asked to look at something else first: how the business actually runs.
Start With How the Work Gets Done
That meant asking some basic questions. How does a new job get quoted? Who approves pricing? Where does customer information live? How does work move from one person or department to the next? Are processes written down, or does everyone just know what to do?
The answers were familiar. Much of how the company worked lived in employees’ heads. Information was spread across spreadsheets, and different people handled the same task in different ways. The business was running, but there wasn’t much structure underneath the work.
That isn’t unusual, and it isn’t a failure. Most growing businesses run this way for years. But it changes what an AI specialist would spend their first months doing. Before building anything useful, they’d have to work out how the business operates, and much of that knowledge isn’t written down anywhere.
In other words, a very expensive hire would likely start by doing process work.
AI Needs Something to Work With
Many owners hope AI will clean up operational problems on its own. In practice, it usually works the other way around. AI can do a lot when processes are clear, information is organized and responsibilities are defined. When they aren’t, you’re asking it to automate confusion.
The research points the same way. A Goldman Sachs survey of 1,256 US small business owners, conducted in early 2026, found that 76% are already using AI, but only 14% are fully integrating it into their core operations. The challenges owners named most often were data privacy and security, a lack of technical expertise, and difficulty choosing the right tools.
That lack of expertise is exactly why hiring a specialist feels like the answer. But an expert can only work with what the business gives them, and a process nobody has defined doesn’t give them much.
Gartner has made a similar point about business information. It predicts that through 2026, organizations will abandon 60% of AI projects that aren’t supported by AI-ready data. That’s a forecast rather than a measured result, so it shouldn’t be read as a precise number for every business. But the direction is clear: the work underneath AI often decides whether the AI succeeds. We wrote more about this in When Your Software Implementation Is Really a Data Project.
Build the Foundation First
So the conversation shifted. Instead of starting with the hire, the focus moved to the foundation: organizing information, documenting processes and making it clear who is responsible for what.
This is also where AI can help right away. Documenting a process used to be one of the slowest parts of this work, and it could take us about five hours for a single process. Now, with the right inputs, it can take one prompt.
Getting ready for AI doesn’t mean putting AI on hold. It means pointing it at the groundwork first.
Once that foundation is in place, the business will be in a much better position to decide what AI should actually do. It may even find it doesn’t need a full-time AI specialist. A few well-built tools on top of clear processes may cover most of what the owner was hoping for.
What It Looks Like When the Foundation Is There
When the structure exists, the results can be significant.
One of our clients needed to match incoming clients with the right staff members. Each match meant reviewing information, understanding the client’s needs and deciding who fit best. The process used to take hours. Today an AI agent handles it for about $1 per run.
That works because the inputs are defined. The agent knows what information to look at, what a good match looks like and where the result goes. Without that structure, the same agent would be guessing.
Not every improvement involves AI, either. Sometimes the right tool is a simple automation. Sometimes the biggest gain comes from deciding who owns a task, and sometimes a step shouldn’t exist at all. Part of building the foundation is figuring out which is which. When AI is the right fit, it often works best as the first pass, so your most experienced people can spend their time on the work that needs them.
Before You Invest in AI, Ask Yourself This
If an AI expert started at your company tomorrow, could you explain how your business operates? Could you show them your processes, where your information lives and how work moves from one step to the next? Or would they spend their first six months figuring that out?
Your answer tells you where to start.
The goal isn’t to have the most AI. It’s to run a business with clear processes, the right people and the right tools. Sometimes that tool is AI.
So instead of starting with “What can AI do for my business?”, start with:
“How should my business operate?”
Once you have that answer, AI has something to work with. And that’s usually where the real savings begin.