What is an AI Transformation?

Transformation means fundamentally changing an organization so that afterwards it is structured differently and works differently than before.

For me, however, transformation also always means that I transform toward the customer – and not only inward.

With AI, we can now do things we couldn't do before, and so the question arises whether AI forces companies into a transformation.

Take AI as an occasion to ask yourselves, as an organization, what this means for you. Let me explain what I see.

For me, five big ideas have crystallized:

• System Administration & Software Development
• Customer Orientation
• Coping with Bureaucracy
• Codifying Meaning
• Marketplaces without Platforms

System Administration & Software Development

By now, everyone has noticed that software development benefits from AI, and everyone is applying it too. This naturally changes the composition of teams and the training of young talent. Business and IT are moving closer together, in exactly the place where it was long overdue.

On the topic of young talent: Of course, companies will continue to need young people who still have to learn! Any culture that does not pass on traditions and knowledge will eventually die out. An organization must constantly renew itself. However, people will learn differently than before.

The topic of AI and education deserves a separate article, so here is just a brief example from my own life.

An Example from My Own Life

I learned all sorts of things about networks during my studies. And forgot all of it again. Now I have used AI to complete tasks in the area of system administration. I have a website that should be found on Google. I have Microsoft Office for my business. And of course I want everything to work professionally: When I send an email via Outlook, it should carry my domain, and when someone googles something I can do, they should find my website.

The AI explained to me how all of this works. But I would never have managed it without the foundation from my studies. I may have forgotten all the details, but I still know all the technical terms and mechanisms one needs to know in order not to be totally overwhelmed by it.

AI is a huge lever in software development and system administration, but without expertise I cannot use it properly. So you still have to learn it – just differently than before!

But I don't want to get stuck on the topic of tech, because at the end of the day every company has customers, and that's what it's all about.

Customer Orientation

Customer orientation is nothing new at all. What is new is that I can now automate "boring" work even better, in order to respond to customers even more individually and with higher quality.

To be able to achieve this kind of transformation, I first need a great deal of basic knowledge about existing methods, all of which contribute to the value delivered to the customer:

• Customer Journey
• UI/UX
• end-to-end responsibility
• agile ways of working
• hypothesis-driven working
• Lean Thinking
• DevOps
• different leadership structures
• and so on and so forth, there is much more

There are organizations that have missed one development or another, or implemented it only half-heartedly. With AI, there is now the opportunity to transform this completely, so that customer orientation becomes even better – or, if I do it right, I even open up new business areas and new customers.

Implementing these methods inevitably leads to having to adapt teams and leadership structures. There are organizations that did this more than ten years ago, and I was part of it. Catching up on this part of the transformation is worthwhile in any case.

This transformation is not primarily focused on AI, because this strand looks toward the customer. But AI can be usefully integrated at every point.

Coping with Bureaucracy

AI is absolutely brilliant at coping with bureaucracy. This opportunity must be seized. Reading and combing through lots of material and filling out and processing forms – for that, AI is a true blessing. It is worth investing time here, because afterwards the AI works so much faster than a human, and you are rid of the tedious work.

This part of the transformation is also so important because it gives you a feeling for where the journey with AI is actually heading. Once you have experienced how AI fills out forms within a very short time – namely after you have given instructions that provide everything the AI needs to fill them out –, you immediately ask yourself:

Why are there still forms for humans at all?

A large area of the AI transformation will be exactly that. At some point, everyone will have their personal AI agent, and when organizations need information from this person that was previously transmitted via a form, in the future the AI agent can do that for the person.

But the AI agent doesn't need a form, and the organization that wants the information continues to work with the information in systems anyway. So why not have the information passed directly from the AI agent into the system, without the intermediate step of a form?

Before this becomes reality, a few questions must first be clarified:

• Identity: How does an AI agent identify itself?
• Security: How do you protect the transfer of information?
• Abuse: How do you detect attacks by "bad" AI agents?

But that this will become part of the future is, in my eyes, beyond dispute.

Codifying Meaning

The moment I use my AI agent to fill out forms for me, it logically comes to mind that other people are using their AI agents too. And even more areas of application come to mind.

For example, I no longer want to have to search through websites where I find nothing. Or where I have to click around for hours before I find something. How nice would it be if my AI agent searched for the best car insurance for me. Can it do that?

No! Not yet today! It could do it, but the insurance company can't give it anything. Today you still have to click through comparison portals, and the AI agent can't do anything with them. On the one hand, because the forms are all built for humans, and on the other hand, because it may not always be clear what a field means.

A huge part of the AI transformation will be codifying meaning.

Okay, that sounds unwieldy, so here are a few examples.

Example: Car Insurance

There are many different car insurance providers. Suppose they all make their information available in a form readable by AI agents. And I am now searching for the best car insurance with my AI agent. Suppose they all have similar prices and similar benefits. Which one should the AI agent recommend to me?

My approach would be: The insurers must codify what they stand for! If I don't watch the advertising, there must be something readable by AI agents that makes the differences clear.

Example: The Piece of Furniture

I am looking for a specific piece of furniture made of a specific wood with specific criteria, and perhaps I also have a few criteria regarding the manufacturer. For the AI agent to be able to find this, the entire meaning of it must be codified and readable by AI agents.

Why don't I say digitized? Or in pure text form? After all, AI agents can read text wonderfully.

That is certainly a good start. But perhaps I want to control much more precisely what the AI agent outputs about my company and my products. That's why, throughout this whole topic, I think of text modules that are put into a deterministic framework, which the AI agent must output exactly as they are. Technically, all of this is already possible today. It's just that hardly anyone has started doing it.

Example: The Independent Consultant

Someone is an independent consultant. This person has a website, a LinkedIn profile, writes a blog on Substack, posts on Instagram and on X, and has written books. With almost everyone I have met so far, the information diverges. If my AI agent now wants to read what the person does and whether I want to hire them, it is pure luck whether the AI agent finds everything about the person.

There is no single source of truth.

You can try to represent everything via the website, but today you still have to touch umpteen places. I think the future will look like this: You will have to codify the meaning – who you are, what you stand for, where you can be found, what you do, and so on and so forth – so that AI agents can read it without any problems.

Everyone will then have their personal "API".

This can be accessed by others, but it also serves internally as a single source of truth and feeds all other digital places with the right information.

I am already right in the middle of building this for myself. In a few years, we will ask ourselves how we ever managed without it.

Marketplaces without Platforms

And if you take this thought even further: Everyone has their codified self, so to speak the part of themselves they want to show – why do we still need platforms then?

Only to be discovered. But how to find others – that is a solvable problem.

Younger people won't be able to follow me here. I have to go back a bit. When the internet came into being, I was already working with computers; I was there live, so to speak, when the internet was made available to the public.

Back then, we had visions and dreams of the internet, of being connected with one another. Over the years, none of these dreams came true. And now I am dreaming them again!

Back then, there was already Schema.org, a kind of code with which you can codify content, and it is partially used for websites. You could already do that back then and connect with one another, but the hurdle was so big:

• the hurdle of creating it in the first place
• the hurdle of keeping it up to date

It was only for geeks. Now, with AI, it is super easy to create something, and I can even make mistakes when codifying – the AI can still piece things together. So it is no longer so extremely difficult.

Now imagine: Everyone has their "API" and offers their services and products there. Or shares their interests. That could become a marketplace. Or a bulletin board.

You could create something new, entirely without the huge platforms we use today.

In companies, it will very soon happen that data leaves the platform, because otherwise the AI agents cannot work across platforms. After all, every end-to-end process toward the customer spans multiple platforms. If AI is to be used here, data must leave the platform, and I must store the meaning of the data – that is what everyone means by semantics, ontology, knowledge, etc.

I think this will also spread to the rest of the world. If I can tell the rest of the world via my API who I am, what I offer, and what I am looking for, then I don't have to limit myself to platforms.

The best example is LinkedIn. The original idea was connecting with one another.

You had contacts. Now I have contacts and followers, and when I post something, it is not shown to them. So I have to adapt to the algorithm in order to communicate something. But I don't do that. So it is shown to no one. After all, I am not a preacher from "Life of Brian" who keeps playing the same record.

If everyone can now connect and communicate with AI agents and we find a way to discover each other, then we can do this without the algorithms of platforms.

Conclusion

Until then, there are still many questions to be clarified, but it is possible. That is my point.

My personal opinion is that this will be the much bigger AI transformation than anything else that is usually talked about. Because the way AI agents will communicate with each other will also have an impact on how companies will change internally.

We are not there yet.

But if you want to get started, then think right away about how you want to tell the outside world – the AI agents of other people – who you are, what you offer, and what you stand for.