Where Should You Start With AI in Your Business?

If you know your business should be using AI but you have no idea where to start, don't start by looking for an AI tool.

Start by looking at your business.

More specifically, look at your workflows and figure out where things are slowing down, where information is getting lost, where mistakes are happening, or where your employees are spending time on work that could be done more efficiently.

Then figure out what would actually make that process better.

Sometimes the answer will be AI. Sometimes it will be an automation. Sometimes it will be software you already have and aren't fully using. And sometimes you may just need a better process.

The point is to figure out what your business actually needs before you start adding technology to it.

Businesses Are Using AI. That Doesn't Mean They're Getting the Most Out of It.

AI adoption isn't really the problem anymore.

McKinsey reported in 2025 that 88% of survey respondents said their organizations were regularly using AI in at least one business function. But only about one-third said their organizations had begun scaling their AI programs, and just 7% said AI was fully scaled across their organizations.

Source: McKinsey, The State of AI in 2025.

So businesses are experimenting with AI. The harder part is figuring out how to make it useful across the business.

IBM found something similar in its 2025 CEO Study. The CEOs surveyed reported that only 25% of AI initiatives had delivered their expected return on investment, while only 16% had scaled across the enterprise.

But one finding stood out to me even more: 64% of those CEOs acknowledged that the fear of falling behind can lead businesses to invest in technology before they clearly understand the value it will provide.

Source: IBM, 2025 CEO Study.

And I think that's exactly where businesses can get into trouble with AI.

There is always a new tool. There is always something everyone is talking about.

That doesn't mean you need it.

Look at How Your Business Actually Works

When I say you should examine your workflows, I mean everything you do to run your business.

How do you acquire a lead?

What happens to that lead's information?

Who follows up?

How do employees communicate with customers?

Where is information stored?

What happens when someone needs to find something?

What has to happen before a project moves from one person to another?

You have to break the work down into the itty-bitty steps that most people don't think about because they're just doing the work.

That's where you start seeing things.

Maybe there's one step that takes far longer than it should.

Maybe people are entering the same information more than once.

Maybe follow-ups are getting missed.

Maybe employees have to search through emails to find information.

Maybe important information lives in someone's head.

Maybe customers are waiting longer than they should because one person has become a bottleneck.

Or maybe there's something the business isn't doing at all because the current process would take too much time.

These are all opportunities to improve the business.

Don't Only Talk to Leadership

This is also why I think you have to talk to the employees who actually do the work.

Leadership knows the business, but they're not necessarily doing every process themselves.

The employees are.

They tend to know what slows them down. They know which task takes forever, which information is difficult to find, which step is annoying, and which process requires them to do the same thing over and over.

But they don't always recognize those things as problems.

If you've done something the same way every day for years, eventually it's just how you do your job.

Someone looking at the entire workflow can ask a different question:

Does it actually have to be done this way?

Don't Assume the Answer Is AI

This is where I think businesses can make a mistake.

They find a problem and immediately start looking for an AI solution.

I don't think that's where you should start.

In businesses I've worked with, one of the most common things I've encountered is disorganization. Information might be on paper, buried in emails, stored in different places or known only by certain people.

Employees end up spending time just trying to find what they need.

If that's the problem, I want to figure out how to organize everything and create a system that brings those pieces together.

Maybe that's a CRM.

Maybe two systems need to be connected.

Maybe an automation would help.

Maybe AI makes sense.

Or maybe the company already owns software that can solve the problem and nobody is using that feature.

If you already have something that works, there's no reason to add another AI tool.

I'd rather look at what you're already paying for, see what it can actually do and teach your employees how to use it. If it doesn't do what the business needs, then we can look for something else.

The Newest AI Tool Isn't Necessarily the Right AI Tool

Just because an AI tool is new and shiny and everyone is talking about it doesn't mean you need it.

First, these tools cost money.

Second, they cost time.

Someone has to learn the tool. Employees may need training. It has to be incorporated into the company's existing processes. And then you have to figure out whether it actually works for your business.

If you don't know why you need the tool in the first place, you could spend a lot of time learning something that doesn't solve a meaningful problem.

Not every tool is for every business.

That's why I would rather start with the business need and work toward the technology than start with the technology and go searching for somewhere to put it.

AI Is More Than ChatGPT

Another thing I see is that when people think about using AI in their business, they often think about ChatGPT, Claude or another generative AI chatbot.

That's understandable. Those are the AI tools most people know.

But that's only the tip of the iceberg.

Imagine you're using a chatbot to write an email. You ask it to draft the email. It doesn't sound like you, so you ask it to rewrite it. You go back and forth until you get what you want. Then you copy it into your email and make a few more changes.

Sometimes that saves time. Sometimes the back-and-forth becomes another process in itself.

A properly configured AI agent could already have instructions about your voice and how you communicate. Depending on how it's connected to your systems and what permissions it has, it could even prepare the email directly in your drafts for you to review.

AI can also be incorporated into recurring workflows and automations rather than requiring someone to start a new conversation every time they want something done.

That doesn't mean your business needs an AI agent either.

It means there are a lot more possibilities than most business owners realize.

And that's another reason I don't think you should choose the technology first.

Look Beyond Saving Time

When you're looking for opportunities, don't only look for repetitive tasks.

Time matters, but it's not the only thing I'm looking for.

I want to know:

Are mistakes happening?

Are follow-ups being missed?

Is information difficult to find?

Are customers waiting unnecessarily?

Is something preventing employees from taking on more important work?

Is the business losing opportunities because of the way a process currently works?

Could improving this process give the company more capacity?

Could that additional capacity ultimately help generate more revenue?

The goal isn't just to make one task faster.

It's to make the business work better.

There's research behind that idea, too. In McKinsey's research on how organizations are creating value with generative AI, redesigning workflows had the largest effect on whether respondents reported an impact on EBIT among 25 organizational attributes the researchers tested.

Yet only 21% of respondents at organizations using generative AI said their organizations had fundamentally redesigned at least some workflows.

Source: McKinsey, The State of AI: How Organizations Are Rewiring to Capture Value.

In other words, simply adding AI to the way work has always been done isn't necessarily the same thing as improving how the work gets done.

Find Everything. Then Pick Somewhere to Start.

Once you examine a business closely, you may find a lot of opportunities.

I wouldn't try to fix everything at once.

Identify the opportunities, prioritize them and start with something that can be done relatively quickly and produce a noticeable benefit.

Let people see that it works.

Then move on to the next opportunity.

That also gives you a chance to learn from what you've implemented rather than changing ten things at the same time and trying to figure out what's working.

And once something has been implemented, go back and check it.

Is the process running smoothly?

Is the work easier?

Are people actually using what you implemented?

Did it solve the problem you were trying to solve?

Implementation isn't the end of the process.

Your Employees Need to Understand the AI Too

You also can't just put AI into a business and expect everybody to know how to use it.

Some employees may already use AI every day.

Some may have never used it.

And some may be afraid of it.

That's why training has to be part of implementation.

If you're changing the way someone does their job, that person needs to understand what the new system does, how to use it and how it fits into the work they're already responsible for.

Technology doesn't help very much if the people who are supposed to use it don't understand it.

So, Where Should You Start With AI?

Look at your workflows and see what's slowing you down.

Start there.

Look at the small steps. Talk to the people actually doing the work. Look for wasted time, missed opportunities, mistakes, scattered information and things that are keeping people from doing more important work.

Then figure out what would actually improve those processes.

Don't choose the solution until you understand what the business needs.

I'm a very analytical person. I love looking at the small details and figuring out how things could work better. I'm also a perfectionist, so perfecting processes is actually fun for me.

That's a big part of what I do through Clarity & Capability's AI Business Audit.

I learn how the business operates. I talk to leadership and the employees doing the work. I break down the workflows, find the places where things could work better, look at what the company already has and identify opportunities they may not have even realized were possible.

Then we can determine what actually makes sense—whether that's improving a process, better utilizing existing software, connecting systems, adding automation, using AI or building something more customized.

Because the question isn't:

“What AI should we add to our business?”

It's:

“What does our business need to work better—and can AI help us do it?”