By Dr. Mike Donaldson

Most small business owners asking how to start with AI are asking the wrong question. The real question is not which tool to pick or which platform to try. It is whether your operation is actually ready to absorb AI and get anything useful out of it. That distinction matters more than any software decision you will make this year.

Here is what I see happen most often. An owner hears about a tool, reads a headline, or takes a vendor call. They sign up. They start. Three weeks later they are stuck.  It is not because the tool is bad, but because they never figured out where they were starting from. The tool was fine. The foundation was not.

Think about painting a room. You go to the store, pick a color, buy the paint, come home, and open the can. Then it starts. The color looks different in your lighting. The brush is too big to cut the corners. Your rollers dried out in the garage. You have no tarps. The furniture is still in the middle of the room. You had enthusiasm. You did not have a starting point.

That is exactly what AI adoption looks like in most small businesses right now. Owners buy the tools before they have answered the basic questions. What problem are we solving? Is our operation ready to hand that task to AI? What does success even look like? Without answers to those questions, buying a tool feels like progress. It is not. It is expensive motion.

There is a statistic worth looking into a little more.   AI adoption among small businesses dropped from 42%  to 28% in one year. The owners did not walk away because AI stopped working. They walked away because they had no clear purpose going in. They chased the tool instead of the problem. Many signed up for three or four tools at once and found themselves paying $300 to $400 a month for software none of them were using well. The financial cost was real. The bigger cost was paralysis.

Before you pick a tool, before you sit through a demo, ask one question: is my operation ready to absorb AI and actually benefit from it?

AI accelerates what already exists in your business, good or bad. A solid process gets faster. A broken process gets faster too and it is a fast broken process. AI does not fix problems. It executes on what you give it. If what you give it is undocumented, inconsistent, or owner-dependent, the output will reflect that.

A useful question before you automate anything: “would I be comfortable if this process ran twice as fast, or 24 hours a day?”  If the answer is any version of no, you have a process problem, not a tools problem. Fix the process first. It is not glamorous. It is not the exciting part of AI adoption. It IS the part that makes everything else work.

The starting point for AI readiness is a 30-minute inventory. Three things to look at honestly.

Your processes. 

AI cannot inherit a process that does not exist on paper. It cannot read your mind, and it cannot read your team’s minds either. If the process is not documented, document it first. Not AI. Process first.

Your pain. 

That is your starting point. The problem that is costing you the most is where AI has the clearest job to do.

Your expectations. 

If the answer is “handle everything” or “fix our problems,” you need a more honest definition. AI handles tasks. It does not handle culture. It does not exercise judgment. It does not build strategy. That is your job. AI executes on what you give it, at the level of clarity you bring to it.

Once the inventory is done, the path forward is straightforward. Not fast. Not exciting. Straightforward.

Step 1: take the inventory. Write down your three most time-consuming recurring tasks. Not the most strategic ones. The ones that consume the most hours. That list is where you begin.

Step 2: ask the honest question. For each task on your list, is the process documented well enough that someone completely new could follow it and complete it to your standard? If not, document it first. That step is not optional.

Step 3: pick one problem. Not three. Not five. One. The most time-consuming task that already has some consistency behind it. One task, one tool, and then you measure it.

Step 4: define success before you start. This is the step most owners miss. How much time should this save? What does good output look like? If you cannot answer those questions before the pilot begins, you are not ready to pilot. You need to know what “done” looks like before you start moving.

The goal is not to be first. It is not to keep pace with whatever your competitors are doing or whatever vendors are telling you. One solid AI process running cleanly three months from now puts you ahead of any owner who bought six tools and abandoned all of them. Knowing how to start with AI for your small business means starting from your current state, not someone else’s benchmark.

Pause before you buy. Walk the room before you paint it. Your starting point is the only one that matters.

By Dr. Mike Donaldson

Ask most small business owners how to start with AI in their small business and you get a version of the same answer. They want to fix the marketing, the scheduling, the reporting, and the customer follow-up, all in the same quarter. They have watched the demos and read the case studies, and they are ready to move fast. That instinct is not wrong. The timing is.

You do not decide on Thursday that you are running a marathon on Saturday. Your body is not ready. You will hurt something, or you will simply fail to finish. Nobody trains that way, and no owner should try to start AI in their small business the same way.

The same rule shows up in the gym with something as simple as adding weight to a bar. You add five pounds at a time until you cannot add five pounds anymore. Then you go smaller. You move to two and a half pound plates, then microplates, a quarter pound at a time if that is what it takes. The increment shrinks as the difficulty grows, but the direction never changes. You keep adding, in whatever size your body can actually absorb that week.

A business works the same way. You cannot go from zero AI use to five automated systems in a month and expect any of them to hold. You have to walk before you add mileage, and add mileage before you add speed. That means picking one process, getting it right, and letting that result tell you what to try next.

I saw this play out on a manufacturing floor years ago, with a Lean 5S initiative instead of AI. A company kicked off Sort, the first S in 5S, and they went hard. They cleared an entire work area fast, pulling out anything that was not clearly in active use that day. The problem was they never slowed down to verify what actually needed to stay. Speed replaced judgment.

Weeks later, people were quietly pulling items back out of the disposal pile because they needed them for jobs that had not come up yet. The initiative lost credibility fast. Morale dropped with it. What should have built momentum instead deflated it, and the team needed a full restart to get the effort moving again.

That is what happens when a good framework gets applied at the wrong size. AI readiness runs on the same principle. A business that tries to automate everything before it has verified what each process actually needs will end up doing exactly what that team did: pulling things back out of the pile because it moved too fast to check first.

If you are still working out how to start with AI in your small business without repeating that mistake, the sequence matters more than the tool.

Write the list first. Put down every process you think AI could touch: invoice processing, appointment scheduling, follow-up emails, inventory counts, whatever fits your operation. Ten items is fine. The list is not the plan. It is just the raw material.

Pick one. Not the flashiest item on the list, not the one that would impress a client. The smallest one, the one with the clearest edges. If you cannot describe exactly where it starts and where it ends, it is not ready yet.

Define what good looks like and what done looks like before you start. Most owners skip this step and find out three weeks in that nobody actually agrees on whether the win happened.

Attach a real, measurable number to it. Not a feeling that things are faster. An actual figure tied to the business, something you can point to later and say this moved because of what we did.

The value of a first small win is not only the process it fixes. It is what you have sitting in your hands once it is finished. You now have a documented sequence for how that piece of the operation actually works, not the version everyone assumes is true, the version that is real. You have a working relationship with whatever tool or system carried the win, so the next decision does not start from zero. You have a real number that shows whether anything actually moved, and that number becomes the baseline you measure the next win against.

That is what makes the second win faster than the first. Not because AI adoption gets easier in general, but because the guesswork is already gone. The businesses that stall out after one win usually treated it as a one-off project instead of the first entry in a growing map of what actually works in their operation.

Lean has a name for the step most owners skip. Plan, Do, Check, Act. Most owners are strong on the first two. They plan the small win, they do the small win, and then they move straight to the next idea on the list. Check and Act never happen.

Check means running a real after action review, what some call a plus delta review. Plus for what actually worked. Delta for what needs to change next time. Fifteen minutes is enough if you ask the two questions honestly. Act means writing both answers down and carrying them into the next win, not just remembering them and hoping they stick.

Every project is a lesson whether you treat it that way or not. Skip the check and the act, and the next AI project runs into the same delays, the same confusion, the same missing pieces the first one had. Nothing was learned because nothing was recorded. Do the review, and the second win starts from a position the first one never had.

The quarter pound plate does not look like progress sitting next to the forty five pound plate. It is progress anyway. The same is true for the smallest AI use case in your business sitting next to the big vision you actually want. Pick the smallest one. Finish it clean. Then add the next increment your business can actually hold.

The businesses getting real results with AI right now are rarely the ones with the biggest plan. They are the ones who picked one small win, finished it clean, and let that become the starting weight for the next one.