The fastest way to get started with AI in your business is to stop dabbling and give one real task two full days, not two hours. Pick something that has sat on your list for a year, something you have never quite got to, and go head-down with AI on it until it is done. Two hours tells you nothing except that AI is confusing. Two days is long enough to get past the awkward part, produce something real, and change how you think about the technology for good. This is the single biggest reason some businesses pull ahead with AI while most stall, and it costs nothing but focus. This is how to get started with AI in your business.
Why most AI efforts stall before they start
Understanding how to get started with AI in your business can make a significant difference in your operations and productivity.
The problem is rarely the technology. An MIT study widely reported in 2025 found that roughly 95 percent of enterprise generative AI pilots were failing to deliver measurable business impact (see this Fortune summary of the MIT findings). The tools work. What fails is the way organisations approach them.
Most people give AI a couple of hours on a quiet afternoon. They type a few questions, get an underwhelming answer, and conclude it is overhyped. That is not a test of the technology. It is a test of how it feels to be a beginner, and beginners quit at exactly the point where the value starts. The businesses that break through are the ones that pushed past that first uncomfortable stretch instead of walking away from it.
The two-day rule, explained
The rule is simple. Choose one meaningful task, block out two days, and stay properly stuck in. Not two hours split across a fortnight. Two days of real focus, backing yourself to be bad at it for a while before it gives you anything back.
Two days matters because the first several hours are spent being lost. You do not know what the tool can do, you do not know how to ask, and you are half convinced you look like an idiot trying. That is normal, and it passes. Somewhere on the far side of that discomfort the work starts to click, and what you build in the back half of those two days is often something you could never have produced the old way.
What two days actually looks like
Here is a real example from our own business. We needed to run a pulse check for a client, an honest evaluation of how their service desk and CRM implementation was really going. We had talked for ages about standardising that kind of work and building some reusable IP around it. It had never happened. It stayed on the list.
So we gave it two days with AI. It started as an Excel spreadsheet, a scorecard of sorts, most of it head-down, backing and checking and going again. It did not stay basic. It ended up genuinely complex, and it properly impressed the client. Two days of investment produced IP we still use today and built real goodwill with that client at the same time, which is a rare combination. Usually you get one or the other. Some of the fundamentals we now rely on every single day were born in those two days.
You have to be prepared to be bad at it for a while
The honest part is that starting is uncomfortable, especially if you are the person others expect to understand technology. Not knowing what a tool is, not knowing how to use it, and worrying you will look foolish trying is an unpleasant place to sit. There is no clever way around it.
It is not the proverbial ten thousand hours, but you do have to give it real time before it rewards you. Leaning in anyway, and accepting that you will be clumsy for the first stretch, is the actual price of entry. Almost everyone who says AI did not work for them simply left before they had paid it.
The question changes after the first real win
Something shifts once a first project genuinely works. The question stops being “can this even help?” and becomes “well, what else is possible?” Could it document our process? Could it draw the diagram? Could it be the helper we actually needed? For us, that question has not stopped since, and it is the reason the pace of change in the business went from occasional to constant.
That is the compounding effect people underestimate. The value of the first two days is not only the scorecard, or the reconciliation, or whatever you build. It is that you now believe, from your own experience rather than a vendor’s pitch, that the next thing is worth attempting too.
How to pick your two-day task
Choose the thing that has been on your list for a year and never got done. It should be real work with a real output, not a toy experiment. The more rules, quirks and fiddly detail it has, the better, because that is exactly where a well briefed AI setup earns its keep. Reconciliations, evaluations, first-draft documents, process maps and scorecards all make strong first candidates.
Then protect the time the way you would protect a client meeting. Two days, properly stuck in. Expect the first one to feel uncomfortable, and judge the exercise only after the second.
Frequently asked questions
How do I get started with AI in my business? Pick one real task that has been stuck on your list, block out two full days, and work on it with AI until it is finished. A short dabble of an hour or two is not enough to get past the beginner stage, which is where most people give up before seeing any value.
Why do most AI projects fail? Not because the technology is weak. Studies suggest the large majority of enterprise AI pilots fail to deliver measurable impact, usually because of how they are approached rather than what they use. Too little commitment, unclear tasks and quitting at the first friction are the common causes.
How long before AI starts to pay off? Sooner than the ten thousand hours myth suggests, but not in an afternoon. In our experience two days of focused effort on one genuine task is enough to produce something real and to change how you see the technology. The return then compounds as you apply it to the next task.
Give your stuck task two days
If there is a task that has sat undone for a year, that is your starting point. We help businesses do exactly this, turning stuck work into working systems with agentic AI and intelligent automation, because we started the same way ourselves. If you want a partner who has already been through the uncomfortable first stretch and come out the other side, book a scoping conversation and bring the task you have been putting off.
