AI in your organisation: where to start and where to stop

Hardly a week goes by without the conversation turning to AI. With customers, with other business owners and, to be honest, right here in our own office too. Everyone can feel that something big is happening. What was still a handy helper for rewriting an email last year is now taking over entire tasks. And the pace is so high that you sometimes don’t know whether to speed up or to hit the brakes for a moment.

What strikes me: the questions I hear are almost always the same. Not the technical questions, but the human ones. Doubt, a little unease and sometimes just curiosity. In this blog I set them out in order: where do you start with AI, how far do you go and how do you keep it safe? Together with what we have learned ourselves. A few pointers instead of ready-made answers.

What do I do with it, and what not?

The choice on offer is overwhelming. Every software supplier now has an AI button, a new tool appears every week and on LinkedIn everyone seems to have finished long ago. It is tempting to start a little bit of everything.

My advice: don’t start with the technology, start with the work. Where in your organisation does a lot of time go on things nobody enjoys? Retyping data, searching through folders, writing the same quotes or reports over and over. That is often where the biggest gain is, and the risk is manageable.

And dare to say no. Not everything has to involve AI. Work where personal contact, craftsmanship or a difficult judgement call is at the heart of it is better supported than replaced. A short list of “this yes, this (not yet)” already brings a lot of calm.

How far should I go?

Perhaps the hardest question. Some want to automate everything, others prefer to keep the door closed for a while. The answer is almost always somewhere in the middle, and it keeps shifting.

What helps: take small steps that you can reverse. Start with AI as an assistant, with a person who always keeps an eye on it and decides. If that works well and you trust the result, you give it a little more room, step by step. That way you build up experience without turning your whole company upside down in one go.

A rule of thumb we use ourselves: the greater the consequences of a mistake, the closer a person stays on top of it. Minutes of a meeting? Fine. A decision about a customer or an invoice? Someone checks that.

What will my customers accept?

A fair question that is often forgotten. Your customers have an opinion about AI too. Some expect you to use it, because it can be faster and cheaper. Others become uneasy at the idea that their data ends up “in an AI”, or that they will soon be talking to a computer instead of to you.

My experience: customers accept a lot, as long as you are open about it and they benefit from it. What they don’t accept is the feeling that something is going on behind their backs. So simply tell them what you do, why you do it and what you deliberately don’t do. And ask them. That conversation is worth more than you think, and often produces new ideas as well.

Is my competitor moving faster?

I hear this fear a lot. You see a post go by about a company that works “fully AI-driven” and you think: I’m falling behind.

Let me be honest: in our experience most organisations are nowhere near as far along as they appear. A few nice demos are a different thing from AI that works reliably in your processes every day. There is a lot of shouting and relatively little actually implemented.

That doesn’t mean you can sit back. The difference will certainly be made in the coming years. Just not by whoever starts first, but by whoever does it well. Better three applications that really work than twenty experiments that vanish into a drawer. Calm and focus are an advantage here, not a disadvantage.

All very nice, but how safe is it?

This is where I always get a bit sharper. Setting up an AI application is surprisingly easy these days. Someone connects a chatbot to the mailbox in an afternoon, a colleague puts a customer file into a free tool to look something up “quickly”. Well meant, but at that moment nobody knows any more where your data is and who can get to it.

Three questions you should ask yourself:

• Where does my data go? Does it stay within your own environment, or does it disappear to a service whose terms nobody has read?
• How are the connections set up? Every connection between AI and your systems is a door. Who holds the key, what may the AI see and what happens if something goes wrong? An AI assistant should never see more than the employee who uses it.
• Does everyone know what has been agreed? The best security falls apart if people don’t know what is and isn’t allowed.

That last one is perhaps the most important. Don’t inform your people with a thick policy document nobody reads, but with a few clear agreements. Which tools do we use? What do you never put into them? Who can you turn to with questions? Repeat it, discuss it in team meetings and make it normal to hesitate for a moment and ask. Do you want to tackle this in a structured way? Then take a look at security awareness training.

My people aren’t as far along as I am. Now what?

Very recognisable, especially if you are enthusiastic yourself as a business owner or manager. You see the possibilities, you have spent hours trying things out and your team looks at you a bit blankly. Or worse: they are afraid their jobs will disappear.

Take that concern seriously. People only come along when they see what it does for them, not what it does for the organisation. Show how AI takes over their most tedious jobs, so there is more time left for the work they are good at.

What works well for us and for our customers:

• Find a few enthusiasts and let them lead the way. Colleagues learn more from each other than from the director.
• Give people room to practise, in a safe environment and without the pressure of immediate results.
• Start with their own work. One example from their own department says more than ten presentations.
• Be patient. Not everyone moves at the same pace, and they don’t have to.

You don’t have to work it out alone

We are living in turbulent times, there is no denying that. But turbulence doesn’t mean you have lost control. With clear choices, small steps, honesty towards your customers and attention for your people, you will get a long way.

At Infodatek Group we are in the middle of this change ourselves. We have been experimenting with AI since 2020, have used it every day in our own software development since 2023 and help customers use it safely, for example with Managed Cowork: the workplace where your whole organisation works safely with AI. With Managed Cowork your data stays within the European Union, as long as you don’t use an AI platform such as ChatGPT, Claude or Copilot. That is how we know what it feels like to hesitate, and also how much it delivers when you get it right. Read more about AI and automation.

Do you recognise yourself in one or more of these questions? Book an introduction: an honest conversation about where you stand and what a logical next step could be.