AI-implementatie

Where do you start with AI in your business?

Marc Cornelius8 October 2026 · 4 min read

Everyone on LinkedIn tells you to try out every new AI tool. Please don't. If you start with a tool, you're tinkering with a process you don't even know yet. Start with the process. Build from there.

The first question I ask

In an intro call my very first question isn't which tool you want, but what your ultimate goal is. Do you want to grow? More revenue? Fewer staff? Less hassle? Better service? Then I ask whether you've written down your processes. Do you have SOPs? What do you explain to a new sales or marketing hire on their first Monday? That answer tells me more about where the gains are than any tool story.

Start at the start, literally

You begin by discovering the whole process, without any tool involved. So grab a plain Excel sheet if you have to. What's step one? What needs to happen in it? What's the next step? And after that? And so on. That's how you map the real process, step by step. Not picking up an existing automation and tinkering with it further, but truly starting over at the beginning. Tinkering with a broken process just gives you a faster broken process. And that's exactly what goes wrong when you start with a tool: you pour money and energy into speeding up something that should have worked differently. Map the whole process first. Then prioritise. And only then decide what AI takes on, what custom software handles, and what a simple automation tool can do.

“Don't tinker with a broken process. Map the whole process first, prioritise, and only then decide what AI takes on.”

How to find the most expensive half hour

To know where to start, I walk along. Watching how it truly goes, looking over shoulders in the sheets, talking to the people who do the work. The most expensive half hour of the week is almost always highly paid staff doing repetitive retyping. At Argo the bottleneck wasn't the expertise but the retyping: putting sizes into Excel, formatting in Word, printing, getting signatures, scanning and emailing again. At Adverge senior people spent hours manually checking tests. Expensive people, boring work. That's where the gold is. The nice part is that people often don't notice this themselves, because it's been that way for years. Just ask someone what the most tedious part of their week is. Nine times out of ten they point straight at the process you're looking for.

Three signs a process is a good fit

So where do you start concretely? Look at three things. It takes a lot of manual work. It eats a lot of time. And nobody enjoys doing it. Tick those three and you've found your starting point. Begin with what hurts most, not with the process that just happens to be easiest to automate.

Why the starting point sometimes disappoints

It can go wrong, so let's be honest about the pitfall. At a marketing agency we took one employee's task as the starting point and put it into a tool to speed up the work. When the tool hit the floor, it turned out that one employee had a very specific approach of their own that didn't fit the tool. Lesson: pick a starting point that represents how the team works, not one exceptional workflow. That's why I'd rather talk to several people than just the boss.

Adoption starts before you build

The nice part: by mapping the process first, you immediately involve the people who'll use it later. That's not a side issue. Appoint an AI Lead who secures internal buy-in, decide together what needs building, and check in after every phase. That way delivery is no longer a surprise but something the team has been waiting months for. Forcing adoption doesn't work. Letting it grow does.

Want us to walk along?

Not sure where to start? That's exactly the question to bring to our table. We map your processes, point out the most expensive half hour and tell you honestly what AI can take on and what you're better off leaving alone. If we see a process isn't a good fit, we'll say so too, even if that means there's less to build. We'd rather have a small starting point that lands than a big plan that sits idle. Without committing you to anything straight away.