Deploying and publishing are not the same thing. Here is how you choose between a web app on a server and a real app in the store, and how you test it on your own device first.

You have built something with your agent and it works. On your own laptop, in your browser, on localhost. And then comes the question that almost everyone in a vibe coding training asks sooner or later: how do I get this to other people? On their laptop, or preferably just on their smartphone.

The first reflex is often the app store. Because an app belongs in a store, you might think. I would take a step back first. There are actually two routes, and they have different names because they really are different. Deploying is putting your app on a server, so people can access it via the web. Publishing is putting your app in a store, like the Apple App Store or Google Play. The sooner you know which route your app needs, the less you have to rebuild afterwards.

Does your chat with your agent run on endlessly during a larger project? Turn it into an orchestrator that distributes the work among other agents, while you keep talking to it.

The larger your project gets, the longer your chat with your agent becomes. In the beginning, everything goes smoothly. After a few hours, it suggests building something it already built half an hour ago. And you are sitting there waiting for it to finish one task before you can start on the next. You might recognise this from your own projects.

In a vibe coding training course, I mentioned that I have been doing this differently for a few weeks now. You no longer let one agent build everything. You turn it into an orchestrator, which distributes the work among other agents. And you only talk to it.

Your app is ready, but setting up a server feels like a separate profession. Give your agent access to your own server and let it propose and handle the technology.

In a vibe coding training, everyone had just seen their first app working on their own laptop. And then a participant asked if he had to set all that up on the server himself. You might recognise this. Building went surprisingly smoothly. And then comes the part most people dread: a server, a domain name, certificates, a database. That sounds like a separate profession, and it actually is.

My answer was short. You say nothing. You let your agent do the work, on a server that belongs to you. In this article you will read how that works, and what you do have to arrange yourself.

Five steps and one central prompt to turn what you already do into a small content machine that writes in your voice. And you remain the one who publishes.

You actually know that you should post more. You also have plenty to say, because you do interesting work every week. Only that work is in your notes, your recordings and your emails, and not on LinkedIn. This step-by-step plan turns what you already do into a small content machine. One central prompt builds it together with you, and you remain the one who publishes. Count on half an hour for the setup.

Start with what is already there

Most people start with the question of what they should post. I would start the other way around: with the places where your content is already being created. A report after a client conversation. A voice memo in the car. An email in which you explain something to someone. The more you look at that, the clearer it becomes that you have plenty of inspiration. What is missing is a fixed route from your work to your reader.

Too many newsletters and still not keeping up? A workflow with a timer searches for news about your field every morning and emails you the three most important stories.

You want to keep up in your field. So you are subscribed to a whole series of newsletters. And every morning they sit there, among your other emails, and you actually do not read most of them. You might recognise this: you do not want to miss anything, and precisely because of that you miss all sorts of things.

In a workshop on automation with n8n, a participant asked if you could approach that differently. He wanted an overview every Monday morning of what was new in his domain over the past week, including the sources. That is possible, and it is one of the most beautiful first workflows I know. You are essentially building your own newsletter, about exactly your topic, and you are the only subscriber.

Does your chatbot answer vaguely or beside the point? Give it a narrow knowledge document, separate instructions and tell it explicitly where to look.

You have built a chatbot for your website or for your colleagues. It answers, which is something. But sometimes it answers just beside the question. Or it tells a whole story that nobody asked for. A participant put it this way in a workshop on automation: 'It does look at my document, but it keeps answering a bit in the wild west. How do I make it smarter?'

You might recognise that. And the first reflex is then to add even more information. I would try the opposite first. In this workshop we built a chatbot together in n8n, and there you saw very nicely where things go wrong.

The assistant in your automation tool knows nothing about your work. Connect the agent on your computer to n8n and let it build, with up-to-date documentation at hand.

You are in n8n and you want a workflow. There is an assistant in the tool, so you just ask him. He builds something, and it looks impressive. But it does not do exactly what you meant. He does not know where your files are, which tools you use or what your week looks like. So you start explaining everything, question after question.

In a workshop on automation with n8n I said that I almost never build my workflows in n8n itself. I let the agent running on my computer do that. That sounded a bit strange to a few people at first. Why would you not use a tool for what it was made for?

A script to fill in for one agenda item in 1-2-4-All, with the sentences you can say word for word and a time schedule. Ready in ten minutes.

You want everyone's input in your next meeting. You also know how it usually goes: you ask if anyone else has an idea, and then the same three people do the talking. The rest nod politely and are already at lunch in their minds. With 1-2-4-All you turn that around. Everyone thinks alone first, then in pairs, then in small groups, and only then together. This template is the script for one agenda item. You fill it in in about ten minutes, and during the meeting you just read out what it says.

What you get

The whole script starts with one sentence, the question you open with. 'Does anyone else have an idea?' is actually a closed question, and no is the easiest answer then. An open question starts with how, when or in what way. Put that question on the screen or on a flipchart, so everyone keeps seeing it.

Is your agent trying to fix the same thing for the tenth time? Open a new session, point them to the stuck session and ask for the root cause.

Your agent has been working on one little problem for twenty minutes. It tries something, it does not work. It says it understands now and tries something else. It fails again. And then it starts over with the first solution. You sit there watching it and you actually already know it is not going to work. Yet with every attempt you hope a little bit that it will work this time.

In the training on building with AI, I called that a dead loop. With the new models, I have encountered it much less since the summer. But it still happens, for example when an agent has to test something in a browser and gets stuck somewhere. It then tries to fix it internally, and that just does not work.

Many good ideas die in a chat. Brainstorm out loud with your AI, in the car or on a walk, and always ask for documents at the end.

How many chats have you had that ended with 'very good idea'? And where nothing ever really came of it afterwards? You had an inspiration, your AI was enthusiastic, you came up with all sorts of things together. And then you closed the window and life went on. The idea is still somewhere in your history, among a hundred other chats.

Usually, that is because of the end of the conversation. Nothing came out of it that you could give to anyone, not even to an agent. In a training session about agents, I explained how I tackle this myself. I often brainstorm out loud with my AI, in the car, and I only stop when documents are ready.

Five prompts to close your meeting with agreements that happen: checking your report, getting actions confirmed and opening the next meeting.

You have probably had a report made by AI before. It looks neat with headings and bullet points, and it is ready before you have finished your coffee. And yet it often remains quiet afterwards. Nobody really knows who was supposed to do what. And at the next meeting the same discussion just starts all over again. So the report was written in half a minute, and everyone waits for each other for three weeks afterwards. Usually, only one thing is actually missing: a name or a date.

This prompt pack helps you to close that loop. It starts at the last minutes of your meeting and runs until the first minutes of the next one. There are five prompts, and they follow the five parts from the article: decisions, actions, timeline, owner and communication. In my training sessions that is called BATEC. It takes you fifteen minutes, and they work in Copilot, ChatGPT, Claude and Gemini.

Testing your own app feels thorough, but you often just follow the ideal path. Let your agent, and ideally a second AI, look for the edge cases.

You have spent a few evenings building your app. Your agent says everything is ready and opens it in your browser. And then you click through it for a bit. The form works, the button works, the overview appears. It actually looks really good. You might recognise that moment, and also the feeling that you can publish now.

In the training on building with AI, I called that vibe checking. It is the sister of vibe coding. You are so in the flow that testing also happens a bit by feel. And then you very easily follow the ideal path, the golden path, as developers call it. What your app is supposed to do, it does. Only you do not yet know what it does when someone tries something else.

Do you lead a meeting and mostly do the talking yourself? Here is how you choose your role moment by moment: expert, coach or facilitator, depending on what the group already knows.

'I actually find it harder to be an expert in a meeting.' A participant said that during the second session of a training course on facilitating meetings. She could not just come up with answers and theories on the spot. She preferred to think at her own pace, and sent an email with all her ideas afterwards. She found facilitating a meeting much easier. 'That requires more people skills than technical knowledge,' she said.

Perhaps you recognise the opposite. You have to lead a meeting, so you prepare your content. You make slides, you list your points and you explain everything. And the others listen. Or they do not listen and check their emails in the meantime. You have then said everything you wanted to say, and yet very little comes out of it.

Are people always a bit lost in your meetings? With the rules, the referee and the scoreboard of a game, everyone knows where they stand.

The meeting starts and things immediately go wrong. Someone's camera does not work. Someone else asks if they needed a camera. And a third person wants to know if they had to prepare anything. You have a beautiful presentation ready, but the first ten minutes go to questions that no one asked beforehand.

You probably recognise this. People are actually always a bit lost. And that is because they do not know the rules. If no one provides the rules, everyone makes their own rules. And those are not always in the interest of the group.

Do the same three people always talk in your meetings? By starting alone, then in pairs, and then in small groups, everyone gets a turn.

At every conference, there is someone who has waited the whole presentation for one moment, the question round. And then he asks a question that makes the whole room go quiet, and not in a good way. Everyone wants to go to the toilet, or have lunch, or just leave. No one wants to watch a discussion between two people for another half hour.

In your own meetings it probably looks a bit smaller, but the mechanism is the same. A few people do all the thinking and the rest just sit there. You ask if anyone else has an idea, and it stays quiet. A participant in a training on facilitating meetings literally told me she ran into this: she asks for input, and from some people it just never comes.

A ten-minute worksheet that runs your own AI idea through the four boxes of the AI Act: four questions, what you arrange per box and a prompt that interviews you.

You have an idea to use AI in your work. And somewhere in the back of your mind is a little voice asking if that is actually allowed. In training sessions I notice that almost everyone knows that little voice. It asks a good question, but it rarely gives an answer. So your idea gets left behind, or you do it anyway, with a little bit of guilt attached. This worksheet runs your idea through the four boxes of the AI Act in about ten minutes. You end up with a box, with what you arrange in that box and with the questions you still need to put to whoever handles privacy at your organisation. You only need your idea, a sheet of paper and your AI tool.

Are you going from meeting to meeting and not getting to your real work? With a small filter, you decide for yourself where your contribution counts.

You open your calendar and there is another new invitation. And your first reaction is probably: oh dear. You already have no time, and now you have to attend that too. I hear that in almost every training on meetings. People slide from one meeting to the next. And they do their real work in the evening, when it is finally quiet.

I once knew someone who was highly frustrated about that. He literally said: 'I am constantly in meetings, I can never do my real work.' And that was important work. Only he also accepted every invitation that came in. When we adjusted that together, it quickly got better. He did not have to be everywhere, just to be there.

Do you have to say you had no time again in the next meeting? This is how you see the open loops in your head, and close them one by one.

In a training session about meetings, I asked the group to raise their hands. Who had ever received a to-do in a meeting, and had to say in the next one: 'I had no time, I have not got round to it yet'? Almost all hands went up. One participant kept his hand neatly down. I congratulated him, because apparently he had it completely under control.

You probably recognise this. And that is not just down to you. Sometimes there are circumstances or systems that hold you back. What I then show in the training is how such a to-do actually works in your head. When you see that, it becomes easier to do something about it.

Ten questions to copy for your notebook in Copilot or Gemini, for a client conversation, your own organisation, a job interview and checking the answers.

You have created a notebook and put a whole stack of documents in it. Reports, a research paper, the notes of a project. And then you stand in front of that empty box, and you do not really know what to ask a stack of knowledge. Usually it becomes 'summarise this'. Then you get a summary that you would have gotten with one document as well, and that is a bit of a shame after all that collecting work.

This swipe file gives you ten questions to copy. They are organised by the moment you need them: a client conversation, your own organisation, a job interview, and checking the answers. It takes you five to ten minutes. And they work just as well in a Copilot notebook as in Gemini Notebooks.

Do you get a long report after every meeting that nobody reads? With five fixed sections, everyone knows what was decided and who is doing what.

You probably know them, the meeting reports that almost nobody reads from start to finish. A few pages long, with who said what, and who reacted to it, and what the counterarguments were. And when you close it after three paragraphs, you still do not know what has actually been decided. Let alone what you have to do by next week.

I see those kinds of reports mostly at political institutions and at some non-profit organisations. Fortunately, companies do it less, but it creeps in everywhere. In a training on meetings, I therefore showed a simple structure. Five sections, in a fixed order. And the last one brings you nicely back to the start of the next meeting.

Afraid your AI is leaking your data? Often the leak is in the share button. Here is how you check what you have shared in the past and what your AI is allowed to do with your conversations.

You read it regularly in the news: 'ChatGPT leaks user images', or 'private conversations with AI pop up on the internet'. And then you might wonder if everything you have ever typed in a chat is floating around somewhere. I would take those kinds of messages with a pinch of salt. There is definitely something going on, only the leak is usually somewhere else than you think.

That came up in a training session, and I explained how it actually works. Because the part you control yourself is bigger than you expect.

At the bottom of this article you will find a prompt to help you check your own settings. Feel free to copy it.

Are you doubting whether you are allowed to do something with AI? The four risk levels of the European AI Act help you place your own use case in a few minutes.

You have an idea to use AI in your work. Having a report of a conversation written out, processing CVs, a chatbot for your customers. And then the doubt sets in. Is that actually allowed? How far can I go? You are certainly not the only one. I get that question in almost every training, and usually with a worried look.

I would say: you need to be aware of it, but you also do not need to completely freak out about it. There is a fairly simple framework that helps you place your use case. That is the European AI Act, which categorises AI according to risk. As you run your own idea through those boxes, you will notice that much of your doubt disappears on its own.

A mandate to fill in that says what your agent is allowed to solve on your server by itself, when, and what it always leaves to you. Ready in fifteen minutes.

You have a website or an app online, and you would like your agent to help keep it up to date. Updates, leaks, certificates, that sort of thing. Only you also want to sleep soundly while it is at work. For that you need a mandate. That is one file that says what your agent is allowed to solve by itself and when. And it also says what it always leaves to you.

This template is that file, with blank spaces. You fill it in in fifteen minutes, or you let your agent interview you first.

Start with the interview

If you do not want to start with a blank page, let your agent ask the questions first. It probably knows your server better than you do by now. But what it is allowed to do, you still decide yourself. And while you answer its questions, you will naturally notice what you had actually never thought about.

Are you barely allowed to do anything with AI at work? Then people will find their own way, and that usually involves their private chatbot. Here is how you prevent that.

Perhaps you recognise this. Your organisation has rolled out Copilot, or its own AI tool. It came with a long list of what is not allowed. And a few people received a licence to 'play' a bit first. The rest are waiting. Meanwhile, every day you see things you could do with AI in fifteen minutes, and which you are now doing by hand.

In a training session, a group told me exactly that. The tool was assigned from above, to a handful of people, and they would decide later what was useful. I honestly said then that I am not comfortable with that. Because if people are not allowed to do anything, they will not wait. They will find their own way.

A short checklist to see where you are tied to one AI supplier, with the prompt that makes your instructions and your context portable.

Many people work with one AI tool today. And that makes sense, because it is easy. But your instructions, your skills and your documents are then often only in that one tool. With this checklist you can see where you are stuck in fifteen minutes. And you immediately see what you can do to switch tomorrow without starting from scratch. You do not need to know anything technical for this.

Start with these three questions

If you only have five minutes, I would ask these three questions. Where are my instructions and skills stored today? Can I download them as plain text? And which task would stop if my tool becomes more expensive or disappears tomorrow? Write down the answers. Then you already know what to tackle first.

Have your AI make slides or a website and everything looks like what everyone else makes. With one file containing your brand style, it looks like yours again.

Perhaps you recognise this. You ask your AI for a few slides for a presentation, and the result looks quite neat. Only you have actually seen those slides before. From a colleague, in a webinar, in just about every video about AI. And the website you built with AI looks very much like the neighbours' one.

There is a simple explanation for that. An AI that knows nothing about your style just chooses the lowest common denominator of all the slides and websites it ever saw. So you get the typical AI look, and that applies to slides just as much as to websites. It is actually a bit ironic that the very tool that was supposed to help you stand out makes you invisible among the rest.

Lots of ideas and no idea where to start? A small project for yourself, without a boss or a deadline, is in my opinion the most fun place to learn to build with agents.

During a training session about agents, someone literally said: 'I have a lot of ideas, but I do not know where to stop.' You might recognise that. You have seen what agents can do, and now it is bubbling. And at the same time, you do not quite dare to start on it at work yet, because there everything has to be right immediately.

There is a place where that is not necessary, and that is at home. A small project for yourself, that nobody is waiting for and nobody has an opinion on. That day I also showed a few of such crazy ideas. They had nothing to do with work. Yet I think they show better than a demo what is possible today.

Do you need to build an agent for a pile of documents? If you mainly want to extract insights from them, a notebook in Copilot or Gemini actually works better.

You probably have a folder full of documents that you would actually like to get more out of. Reports, research, notes from a project that has been running for months. And perhaps you are already wondering if you should build an agent for that. That question also came up in this week's Copilot workshop. Should I make an agent for that now, or a notebook? Almost everyone mixes that up.

My answer is actually quite simple. If it is purely about unlocking information, a notebook is more convenient than an agent. An agent is meant to do things for you. A notebook lets you ask questions to a pile of knowledge, and it does not need to do anything more.

Anyone who wants to automate usually thinks of the biggest process right away. I would start with the small tasks that you do the exact same way every time.

When we get to building skills in a training session, I always ask what people would like to automate. And then I almost always get the same kind of answer. Someone names the biggest process in their job. The work they have dreamed of for years that would one day happen by itself. Usually, that is also the work that involves the most thinking.

I understand that. Yet I always say the same thing: put that out of your mind for a moment. You do not start there. Look at the low hanging fruit first, the little things you do again every week.

You often do not really know how you sound. Your AI can extract it from your own emails, and then it will write a bit more like you from now on.

In this week's Copilot workshop I showed the settings for emails in Outlook. There you can give Copilot fixed instructions for every email it drafts for you. As a joke, I had put in that it must open every email with 'Your Highness'. And the day before I had noticed that it actually does that.

There was laughter. But then came the real question. What do you actually put in there? Most people want their emails to sound like themselves. Only they do not really know how they sound when they write. And that is difficult too, because you see your own style the least.

Six prompts to copy that take you from a briefing to a built screen that really looks like your design, with a description that you read through yourself first.

Creating a design with AI actually works very well today. It usually goes wrong the moment your agent has to build that design. Then you get screens that work, but that look different. That is why I simply listed the prompts that I use myself, from the briefing to the built screen. The first two are for Google Stitch or another design tool. The others work in any agent that can read images. Count on fifteen minutes per screen.

What you get

You get six prompts, in the order in which I use them. Two are for the design tool, and four are for your agent: describe, adjust, build and compare. You already get the first one here. It lets your design tool explore three directions instead of one, so I think you see much faster what appeals to you.

An agent that patches your server itself needs a mandate. This is how you decide which sources it believes, what it is allowed to do itself and what it leaves to you.

An agent that monitors and updates your server itself sounds reassuring. Until you wonder what it is allowed to touch while you are asleep. In the previous article you read how Tom Verheyden forwarded me a warning about new security leaks. And how my Webmaster agent then built two skills and an automation in about twenty-five minutes, which run twice a day.

Here you can see how that works, so you can adopt the principle for what you have online yourself. I will leave out my own settings. Copy the starting prompt at the bottom of this article, and let your agent make an inventory of your own server first.

An agent keeps working until it is done, if you tell it what done means. This is how I describe in advance what a good result looks like, as a checklist.

There comes a moment in my training sessions when the foundation is ready. The folders are set up, the first skills are built. And then I ask a new question: what do you want your agent to do while you go and eat? Then it suddenly becomes about something else than asking a question and getting an answer. It becomes about an agent that continues working on its own.

There is a word for that, loop engineering. You are going to hear that a lot in the coming period. It sounds technical. But the idea behind it is simple, and I use it very often myself.

Did you put something online yourself after a workshop? After a week full of security leaks, you can read here how an agent guards your server for you, including the prompt.

Last night Tom Verheyden forwarded me an email from his hosting provider. It warned about a series of security leaks that became known this week. It was about even more security risks related to PHP, about Node.js and about eMudhra, a certificate authority whose keys were stolen. Tom added one line himself:

'Hmmm... it does not seem to stop with just PHP.'

With a half-awake head I read that this morning over my coffee. I was wide awake immediately, and that was not because of the coffee. My first thought went to you. Perhaps you put something online yourself after a workshop on Claude, AI Agents or Vibe Coding: a small tool, a little app, a website. Then this news is actually about you too.

Building a skill sounds like a lot of writing. Today I leave that to my AI. Four ways to let your AI create a skill, including by demonstrating it.

When I started with skills, I wrote them out completely myself. Every step, every exception, in a Markdown file. I actually do not do that anymore. You just ask ChatGPT, Claude or Copilot to write a skill, and it knows how to do that. A folder with a lowercase name, a SKILL.md file in it, the right description at the top. Your AI handles that itself.

What you still do yourself is tell your AI what the skill has to do. And there are different ways to do that. This week I put them side by side in a training, because many people think there is only one.

Before I build a skill or an agent, I look at the shape of my work. A chain, a branch, a nest or a diamond, and what that changes.

When I ask people to describe a workflow, I almost always get a list. First I do this, then that, and then that. That is just how we think. This week I saw it again in a training, when we started building skills. Everyone wrote their steps neatly below each other.

That is actually a good start. But a lot of work does not run in a straight line. And when you see the shape of your work, you also know better what to ask your AI. So in that training I showed a few shapes, with examples from my own work.

At the bottom of this article is a prompt you can copy to see the shape of your own work.

Six steps with ready-to-use prompts to build an agent that goes through your emails every morning and tells you what needs your attention today. Ready in half an hour.

An agent that goes through your emails every morning is, in my opinion, the best first agent you can build. It is small. And you will use it the very next day. This step-by-step plan takes you from an empty chat to an agent you have tested yourself. I wrote it for Copilot, but the prompts also work in Claude or ChatGPT. Count on half an hour.

The first step, in a regular chat

I actually ignore the agent builder. I just start in a chat. And then I let my AI ask questions first. The vague parts in your assignment are exactly what it needs to figure out. Fill in the part in brackets.

The big AI companies are bringing you in cheaply right now. I prefer to spread my work across multiple tools, and I explain why and how you can do that yourself.

In a training about Copilot this week, we talked about licences. Microsoft is changing a few things again, and new features are being added. I also gave my own opinion then, because there is a question behind it that many companies are not asking today. What happens if you put all your work with one AI provider?

I do not tie myself to that. And that has little to do with Microsoft alone. It applies to all the big players.

At the bottom of this article is a prompt you can copy to see where you are dependent.

Who is paying today

A skill is a folder, and it can contain more than just text. This is how I let my agent look at an interesting skill first and make its own version of it.

In a workshop on vibe coding this week, someone asked what a skill actually is. I thought I had already explained it. But I give so many training sessions that I was not sure myself anymore. So I just explained it again. A skill is a step by step description of how an agent should do a task. You can see it as a fixed way of working, in plain human language.

And then the next question usually follows immediately. There are thousands of skills online, and someone asked if you cannot just download them. I stay away from those myself. And I would like to explain why.

An agent that has to build a design often turns it into something else. I let it describe the design first, and only build it afterwards.

Last weekend I rebuilt my website, with AI. I absolutely did not want it to look like all those typical AI websites. So I went through all the steps that I also show in my training courses. First brainstorming, then designing, and only then building.

Along the way I got a new idea. People also had to be able to register on my website, for example to make an appointment. And then I started thinking further about what else they could do there, once they are registered anyway. That is how a complete learning space was added. And for that I first took a step back, to the design.

Six fill-in-the-blank messages that tell your AI who you are and how you work, plus the question to check what sticks afterwards. Done in fifteen minutes.

You know your AI would work better if it knew you. Except you never get round to it, because you do not quite know where to start. This template takes that doubt away. You get six short messages with blanks, in the order you would get to know a new colleague. Count on fifteen minutes and one conversation.

Start with the first message

Every message starts with 'Remember this about me'. That way your assistant knows it has to save something and not answer a question. Fill in the part in brackets. Short is fine, but honest is important.

Anyone who builds tools themselves with AI will sooner or later run into IT. Why this resistance exists, and why a small, concrete request works much better.

On the second day of a vibe coding training, I first ask everyone to share what they have done in the meantime. One participant had put vibe coding on the agenda of a meeting with the IT team that week. That was not well received, there was some serious opposition. Another participant also did not get any further at work. They just built a website about music instead, just to be able to learn.

I actually hear that in every group. And I understand the people in IT too. So I want to explain how I see it, and what I would do.

AI feels almost free today, because someone else is paying the bill. I would already work as if you are paying the real price.

During a Copilot workshop, we had been asking all sorts of things all morning. We translated texts, had summaries made and tried things out. I had typed 'remember about me that' a few times in a row myself, just to show how that works. And nobody stopped to think about what all that costs. I think that makes sense, because it feels like it costs nothing. So I wanted to talk about tokenomics for a moment. You are going to hear that word a lot in the coming years.

At the bottom of this article is a prompt to summarise your longest chat, so you can continue in a new conversation.

An agent that checks your emails every morning is an ideal first agent. This is how you write the instruction and how you give it only what it needs.

I am very often on the road to give training myself. And then I do not read my emails, of course. And I am far from the only one in that, as it turned out during the last workshop on Copilot. During the introductions, everyone talks about their mailbox, which is always overflowing. One person literally said: 'I have already tried everything: rules, folders, categories, ... but I get hundreds of emails a day! It is simply impossible to keep up with!'

And then I demonstrated how you build an agent that checks your emails. And it does one thing: every morning at 06:00 it checks your emails from the previous day, and gives you a summary: what is a priority, what is spam, which actions you definitely need to do today.

If you are translating into a less familiar language or about something sensitive, I would ask your AI how certain it is first, and then translate it back with a second AI.

When you have AI translate a text, you almost always get a fluent translation back. And because of that, you do not really see if it is actually correct. We talked about that in a workshop. A participant from the care sector said that she works with languages every day. Those include languages from the Horn of Africa, and you do not easily find a colleague there who can just read along. She had never used AI for that. She thought the knowledge of such a language would be limited. That might well be true, but I told her that you can simply ask your AI about that.

A fifteen-minute worksheet that sorts your documents into three buckets: leave as is, business or European, and local. Includes the prompt that helps you sort.

You roughly know what you have put into AI over the past month. A quote here, a report there, an email from a client you wanted to rewrite quickly. But roughly is not an answer when someone asks where that client's data has been. This worksheet puts on paper in fifteen minutes what you use where, and what you should do elsewhere from now on. You only need your AI tool, a piece of paper and an honest look at your history.

Three questions per document

Take one document you recently put into AI and ask three questions about it. Does it contain names, addresses or other data about people? Is it from a client, or did a client entrust it to you? And would you find it awkward if someone outside your company read it? Three times no, then it can stay where it is. One time yes, then it belongs in one of the other buckets. Below is the complete worksheet, with the four steps and the prompt that helps you sort without having to paste a single sensitive document.

No request gets left behind and no follow-up gets forgotten. How AI helps you answer quickly and follow up on time, without leaving your clients out in the cold.

You spent a whole day with a client. In the evening you look in your inbox and see a request that came in early this morning. You answer immediately, but you already know. Anyone who asks a question in the morning and only gets an answer after dinner has already emailed another provider in the meantime.

I think that has little to do with carelessness. A request that gets left behind is rarely a bad request. Usually your day was just too full. And the fuller your days become, the more often this happens, and the less time you have to notice it.

The manual work that comes back every week does not have to stay with you. What AI can take over from your emails, reports and quotes, and how you start small.

You are sitting at the kitchen table on Sunday evening with your laptop open, when you actually wanted to be on the sofa already. Quotes still need to go out. Three clients are waiting for last week's report. And your inbox is full of requests that you have to answer just a little differently each time. It is usually not difficult work. It does come back every week, and it costs you exactly the hours you wanted to give to your clients.

You might recognise this. I see it with many freelancers I meet. They are good at their profession, and yet they are stuck with the manual work around it in the evenings. You are probably already wondering if that can ever change.

One card to hang next to your screen or share with your team, plus six sentences to adjust your AI in plain language while it builds for you.

In my building training sessions, one sentence comes back so often that participants finish it themselves after an hour. Dumb it down, chunk it down, and babysit your AI. It fits on a card, and you can find that card below. Hang it next to your screen, put it in your team channel, or send it to that colleague who just told their agent to 'solve it yourself'.

The card

You are free to share it, as long as the name stays on it.

Frequently asked questions

What does dumb it down, chunk it down and babysit your AI mean?

Dumb it down is saying it in plain language, like to a smart colleague who does not know your project. Chunk it down is cutting your project into small, checkable pieces. Babysit is going to look if it is correct after every step before you ask for the next one.

A new colleague gets an onboarding, your AI usually does not. I use six questions from my coaching practice to teach my assistant how I work.

In a Copilot workshop I asked the group to type one sentence: tell me what you know about me. Not much actually came out of that. A job title, a few recent files, sometimes nothing at all. Then I showed what my own assistant has remembered about me. My way of working, what I find important, who I work for and where I want to go. That had nothing to do with the licence. I just did one exercise once: I gave my AI an onboarding.

Below this article you will find the template to onboard your AI in six messages, free for Members.

Crossed arms or looking away say little on their own. I first look at how someone normally is, and then I pay attention to what deviates from that.

I once sat at a table with the CEO of a company where I was going to give a leadership programme. The conversation with HR had gone smoothly. Then he came in, clearly stressed. He sat down, opened a folder and said: go ahead and tell me. While I was talking, his eyes constantly darted in all directions. After a few minutes, he closed the folder. 'I know enough.'

I knew at that moment that the conversation was not going well, and I also knew why.

At the bottom of this article you will find a prompt with which you can have AI make a small observation sheet for you to practise this.

Your advantage lies in working more personally. AI helps you work faster and more tailored to your client, with your own voice included.

After a training session, all participants usually get the same email. Thank you for attending, here are the slides, see you next time. That is polite and everyone does it that way. It is almost funny: the more polite and general the email, the faster you forget it.

Think about an email that is different for each participant. One that refers to what that person said themselves during the day. And that asks what they can move forward with now. In terms of content, that is a small step. In the past, you simply could not take that step, because you did not have the time for it.

You rarely solve an inner conflict by choosing. With the Visual Squash you look for what both sides of yourself want, and that is often the same thing.

A participant stood at the front with her hands open in front of her. I asked her to describe the two sides in her own words. In one hand she placed the part of herself that gives a great deal to her family. In the other hand she placed the part that wants to do everything itself and struggles to let go. Then I asked her to slowly bring her hands together. Halfway there she stopped. Her arms tensed up and her fingers trembled. 'That clashes', she said.

I then said: 'Okay, that is the first step.' You recognise the two sides, and you feel that they clash.

Many small costs keep running because nobody looks at them. Which tasks you can take into your own hands with AI, and how you start small without stacking subscriptions.

At the end of the month you sit with your bank statements. There is a subscription, an invoice from the freelancer and the virtual assistant. Next to that you see the little tool that types out your conversations and another tool to schedule your social media. Each amount on its own is not too bad. Together they form a row that you would rather not look at for too long.

Ask yourself what you are actually paying for. Sometimes it is a profession that someone else can do better. Sometimes you just never got around to it yourself. You probably recognise this, and perhaps you are already wondering which of those expenses belong to the second type.

A short checklist for anyone building apps with AI, including the security paragraph for your brief and the prompt to have a second agent attack your app.

Building an app with AI goes so fast today that security easily becomes an afterthought. And later rarely comes. This checklist fits on one screen and walks with you through three moments: when you write your brief, while you build, and right before you go online. You do not need to be a developer to use it. You just have to ask the right things.

The three questions that catch the most

If you only have five minutes, start here. Where are my passwords and keys, and do they go online? What happens if someone pastes malicious text into my input fields? And where is the data of my users stored, in Europe or outside of it? Ask your agent those three questions and really read the answer.

If your tool is not in the list of connectors, your AI can often build one itself. This is how I tackle that safely.

In every workshop I open the list of connectors at a certain point. Those are the connections between your AI and your other tools: your mailbox, your calendar, your design programme, your CRM. And then you see the group change immediately. Someone scrolls along and sees dozens of logos. And she wants to click them all, because imagine you might need something one day.

I would not do that. Every connector that is turned on tells your AI what it can do during every conversation. That takes up space in its working memory, even if you do not use that tool today. So it is best to only turn on what you actually use. Do you only need a design tool occasionally? Then you turn it on when the time comes, and simply turn it off again afterwards. It is not going anywhere, you know.

An expensive personal AI subscription does not give you business agreements about your data. I explain the difference, and what I keep local myself.

In a workshop about agents, a question came up that I would actually like to hear in every group. A participant works as an external consultant. And they put contracts into AI, with an expensive personal subscription. So they wanted to know if anyone else can access that, or if it is protected.

My answer surprises most people. A personal subscription remains personal, no matter how expensive it is.

Below this article is a worksheet about which documents can go into which AI, free if you have an account.

Personal or business

Dutch costs your AI more working memory than English. So I let my AI work in English, and one sentence in your instruction is enough for that.

In the workshop I sometimes show the instruction of my own agent. People expect something complicated then. What stands out to them most is one little sentence at the top: you do all the work in English. You think in English and you write your intermediate steps in English. And you brief other agents in English too.

And then comes the question I see coming: 'Steff, you write everything in Dutch, right?' That is true. My articles and my emails are in Dutch, and my workshops too. Only what my AI does along the way to get there happens in English. And there is a very practical reason for that.

You repair a car by looking for the cause, not a team. I explain why that difference matters and where we often go wrong.

A few years ago we were driving back from France. And then the lights on our dashboard started to come on. Orange and red lights, usually not a good sign. It was Sunday and we still had a few hundred kilometres to go. We did get home. But we needed someone who knew about these things. And the mechanic did what a good mechanic does: he searched. First in the engine, then in the parts and then in the parts of those parts. Until he found a small filter. He replaced the filter and all the lights went out.

I tell that story in a workshop on strategic thinking, because it fits so beautifully. There is one cause, you find it and you solve it. I think we are trained to think like that. And that is exactly why things go wrong when we start thinking like that about something that is not a car.

Ten sentences to copy for the start, the middle and the end of a long AI task, so you work with a handover instead of a summary of a summary.

A long task with AI rarely goes wrong in a single moment. It slips away. An agreement about the tone disappears, an exception is forgotten, and three sessions later your AI is working on a project that is just not yours anymore. With the right sentence at the right time, you stop that. These are the ten I use myself, ordered by the moment you need them.

How to use this

A swipe file is a cheat sheet. You do not read it from front to back, you grab the sentence you need right now. Save this page to your favourites, or paste the sentences into a note next to your AI tool.

A SMART goal sounds solid, and yet nothing often comes of it. What is missing when a goal only comes from your head, and how I make it SMARTER.

Recently an entrepreneur told me about his dream. He wanted to build a place where young entrepreneurs with an idea could go. There they would get guidance and a network, with all the trimmings. I said: 'That is an incubator, that is actually super cool.' So I asked what was holding him back. And he said: 'I would have to win the lottery first.'

I looked at him in surprise. And I asked what his business plan was for those young entrepreneurs then. All play the lottery together? It sounds like a joke, but it touches on something I see in almost every workshop on goals. We formulate wishes, and we call them goals.

Deleting one line in Excel cost an organisation its collaboration. Why lists hide what is going on, and how you learn to draw arrows.

Have a look at your own notes from the last meeting. Chances are it is a list. A few points with asterisks in front of them, something you scribbled next to it later, and then another asterisk. In a workshop on strategic thinking I predict to the group what their notes will look like later. And that is usually right.

That is actually how we were raised. We are trained to think in lists. I do not mind that, until you make decisions based on that list. A list does show what is there. What relates to what, you do not see in it.

You can give an agent a goal and let it go, or watch along step by step. I teach the second in my workshops, and I explain why.

There are two ways to build something with AI. In the beginning they actually look the same. With the first you write a good, detailed description of what you want. You give your agent a goal, and then you let it go. With the second you work step by step. After each step you look at what it has done. You try to understand it, and then you adjust.

In my workshops I teach the second. The models can do a lot, and they get better every month. But I have seen what happens when no one watches along anymore.

Below this article you will find the share card with the babysitting loop, which is free for Members.

Anyone who puts all their documents into a knowledge base for AI will see their assistant drown in it. A short index that points the way actually works much better.

Last week in a workshop about agents I showed something I normally keep for another training course. I showed the second brain that my agents use. A participant immediately asked what the difference is compared to a knowledge base where you put all your knowledge. I thought that was a good question. I actually know companies that are doing exactly that. They put their entire knowledge management into one large knowledge base, so that AI can access it everywhere. That is a logical thought. And it is also the mistake I made myself.

I used to build a separate agent for every task, until I no longer knew who did what. Now I work with a few agents that share skills.

In a workshop about agents I got a question that took me a long time to figure out myself. One agent can have multiple skills. So why not just build multiple agents that each do one thing well? That seemed a lot clearer to that person.

That is actually how we did it too, about eight months ago. We built a separate agent for every task. One for translations, one for newsletters, one for quotes, one for reports. Each with its own instructions and its own conversations. That felt very organised at first.

At the bottom of this article is a prompt that helps you tidy up your own agents, which you can just copy.

Six prompts to copy that make your AI ask the right questions first, for an email, a plan, a doubt, a difficult conversation or your own assistant.

You might already know the principle: your AI writes better if it is allowed to ask what it is missing first. Only, on a busy Tuesday afternoon, you have little desire to come up with that prompt yourself every time. That is why I listed the six I use most often myself. Copy, fill in the brackets, paste.

What you get

They are six prompts for situations you probably recognise. Think of a sensitive email, or a plan you need to write out. There is also a text for people you do not know well. And there is a prompt for a choice you are doubting about. The last two are about a difficult conversation and about the instructions for your own assistant. They all end with the same reversal: ask first, write later. They simply work in all common tools, from ChatGPT to Gemini.

Anyone building with AI will sooner or later be asked if they can commit. I explain what Git, a repo and GitHub are, without the jargon.

After a few hours of building with their AI, almost every group reaches the same moment. The agent then asks if it can commit the changes, and no one understands what it means. And then everyone looks at me. The word sounds like something you need training for.

That is actually not too bad. Anyone who has ever saved a Word document already understands the most important part of Git. Only the jargon holds people back. And that barrier is lower than it seems.

Feel free to copy the prompt at the bottom of this article, and your AI will set up version control for you.

ChatGPT, Claude, VS Code, Cursor, Craft Agents: beginners often lump them all together. I work with three layers, and they help you choose.

At the end of a long workshop day, I asked the group what they took away from it. And one participant said something I have not forgotten since. He had used two programmes side by side all day. And only now did he realise they were two different things. Another participant did not see the added value of an agent app, because 'I already have ChatGPT'.

Those are actually questions everyone has. Only not everyone dares to ask them. And they all have the same cause. We talk about AI tools as if they are one thing. But actually there are three, slotted neatly into each other.

I now brief my AI with five letters: task, outcome, unknowns, context and handover, because today's models guess the rest themselves.

In a workshop last week, I built an onboarding assistant for an HR department live. I did not write a half-page prompt for it. There was no role in it and no list of tone instructions. It was five short steps. And I did not upload a single document for it.

Copy the prompt at the bottom of this article, and you will have those five steps ready for your own task right away.

Why the SuperPrompt has done its job

I have worked long and often with an extensive prompt framework myself, CRAFTER. With the older models, you had to write everything out. They did literally what it said. And they filled in the rest themselves, often incorrectly.

Our feeling is already there before we start thinking. I explain how that works, and why emotional awareness actually starts in your body.

Last week in a workshop on emotional intelligence, I talked about buying a house. You go to visit it and you stand in the living room. And somewhere inside you, it thinks: I am going to feel good here. At that moment, the decision has actually already been made. The calculation only comes later.

At the bottom of this article is a prompt that helps you notice where a feeling is located in your body.

The illusion of the rational human

We like to live in the belief that we are rational beings. We think we are constantly making conscious choices. But that is not actually true. What ends up in your shopping trolley is largely impulse. And the biggest financial decisions in a human life are precisely the ones most driven by emotion. I am thinking of the house and the car. If you were to buy a car purely on features, it is perfectly comparable to others. And yet you buy that specific one, because it looks nice.

When my AI made a mistake, it started defending itself. I asked it for a lesson, and that is now in its memory.

This summer I came back from a trip. And the Monday after I had to give a training. During my holiday I had not really followed my emails. Had the confirmation come in or not? So I asked my AI assistant. It went looking and said it had found nothing. I was already standing at the client's door by then. And I really started to sweat.

Feel free to copy the prompt at the bottom of this article, you can use it the next time your AI misses something.

The same question to a second assistant

I thought to myself that this could not be right. So I asked exactly the same question to a second assistant, a different model. It found the email within seconds. It was just sitting there.

If you repeat someone's words in your own words, the connection breaks for a moment. Why I reflect things literally, even in an age of chatbots.

This week I gave a workshop on emotional intelligence. In it, I did a short demonstration with a participant. Afterwards, we discussed with the group what had happened. Most had seen that I adjusted my posture. What stood out less was actually the most important thing. I had used her words, literally.

At the bottom of this article you will find a prompt to check which words you adopted after a conversation.

Paraphrasing feels like listening

We all learn that listening well means summarising what the other person says. And you do that in your own words. That is called paraphrasing, and it seems respectful. You show that you have understood it, and that you have even processed it.

Two colons and a short word, and your AI knows what you mean. This is how I put custom commands in my custom instructions, for what I ask every week.

In a workshop this week I showed my custom instructions on the screen. They state that I do not want to see emoji. And not too many bullet points, and normal capital letters. And below that is a list of lines that start with two colons. Those are the rules I use most often, and they are barely a word long.

Feel free to copy the prompt at the bottom of this article, then you can write your first shortcut together with your AI.

What you type again every week

Have a look at your own chat history. Chances are you formulate the same questions over and over again. Translate this to Dutch. Write a short bio about me. Make this text a bit looser. Every time you type a sentence, sometimes two. And every time it is slightly different, so the result is too.

Not every task needs a skill. I would create one for work you do often, and only at the end of a good chat.

Should I create a skill to choose a new washing machine? I get that question in some shape or form in almost every workshop. Since everyone knows what skills are, it keeps coming back. And my counter question is always the same: how many washing machines are you going to buy? Maybe one every ten years, if things go well. So I would not create a skill for that.

For the tasks where it does pay off, you will find the question I ask at the bottom of this article.

The washing machine test

A skill is a fixed method that your AI can call upon. It contains the steps, the quality criteria, the tone and the format. You write it once. And after that, your AI follows the same path every time. That is where its value lies, and also its limit.

One assistant who knows everything and is allowed to do everything loses track. That is why I work with smaller agents, each with its own defined context.

Think of a small business with one employee who does everything. The accounting, the marketing, the customer service, the planning and the product development. That might work for a while. And then it goes wrong anyway. Everything gets mixed up, nothing is quite right, and their head overflows.

I see many people setting up their AI exactly like this. They have one assistant and one conversation. And they throw all knowledge and all tasks into one pile there.

If you want to divide your own questions into domains, you can copy the prompt at the bottom of this article right away.

An AI on your own computer can read, organise and edit files. I explain what that changes, and how you start safely.

In a workshop we gave the AI access to one folder on a laptop. And then we looked together at what happened. First a messy download folder was tidied up. After that loose Excel files were merged into one overview. And at the end a first version of a follow-up email was ready. It was the same AI as always, only it no longer ran in the browser. I literally said then: 'You actually become much more a manager of work than an executor of work.'

If you want to try it yourself, copy the prompt at the bottom of this article and give it to your desktop app.

Anyone building with an AI agent has to explain where everything is over and over again. One Markdown file solves that, and I will tell you what belongs in it.

When I open a project in a building training, participants usually look at the code first. And then I point to another file. A plain text file, a few screens long, without a single line of code. That file determines whether your agent can work independently, or if it comes to ask you something every five minutes.

At the bottom of this article is a prompt that lets your agent help you draft such a file.

The problem everyone knows

If you work on a project with an AI agent for a while, you know the feeling. Every new session starts with an explanation. This is the project and this is how it is put together. The files are over there, and this is how we want it. If you forget something, the agent goes looking for it itself. And then it makes assumptions that you have to correct later.

Your workflow works today and might break next month after an update. That is why understanding how AI works pays off more than memorising buttons.

Perhaps you recognise this: you have just got a workflow running, and a few weeks later it works differently. I sometimes start a workshop with the sentence: 'Everything I told you before the summer is in the past.' I say that half jokingly, half seriously. The AI world has completely changed again in a few months. And fortunately the principles remain the same, so you can build on that.

Copy the prompt at the bottom of this article. With it, you test whether you understand what an AI function actually does.

What happens after an update

Building an app with AI is fast. If you want it to be secure too, you put security in your briefing and let a second agent attack your work.

Some vibe coders tell me they have built three apps today. Then they must be one hundred percent sure that those apps are also secure. I say that in every building training. And then I always see a few people swallow hard. Making an app work has actually become the easy part. You do not get a secure app automatically.

Below this article you will find the checklist for building securely with AI, free for members.

Why afterwards does not work

The classic reflex is: build first, and if it works, we look at security. If you build with AI, that is extra risky. The agent makes choices about the architecture at every step, based on what it knows. If it does not know that security is important, it chooses the fastest route. And that fastest route is often difficult to change afterwards.

Can AI handle the reservations of a B&B? I always start with a drawing of the process, because that is what actually determines whether an agent works.

Someone I know started a B&B this summer. And they asked me: 'Can that not be done with AI? So that I do not have to check it?' The short answer is yes. But I said first: describe how a reservation like that happens. I have liked using that example in workshops ever since. And it actually always starts with a drawing.

At the bottom of this article you will find a prompt that lets AI look at your process drawing, which you can just copy.

What is really behind a reservation

When we drew it out together, quite a lot came up. Bookings come in via a booking platform, via email and via text message. And sometimes also via comments under posts on social media. Each channel requires its own access, its own settings and its own login details. An agent obviously cannot reply to a message it does not see.

Sometimes Copilot seems to know everything, and sometimes it cannot find a document that is right there. I explain how its index works and what you can do about it.

The same thing happens in almost every Copilot workshop. Someone asks a question about a project and gets a surprisingly complete answer. It includes references to emails and a Teams conversation. A document on SharePoint is also included. A few minutes later, a colleague asks about a procedure that has existed for years. And then Copilot knows nothing. That confuses people, and I understand that.

At the bottom of this article is a prompt you can use to test for yourself which of your documents Copilot finds.

How Copilot gets its knowledge

Those who use AI as a search engine get confident nonsense and think less for themselves. Those who use AI as a thinking partner become sharper. The difference lies in the direction.

'If you use AI in a stupid way, you become stupider. If you use it in a smart way, you become smarter.' I say that in almost every workshop. Usually, there is laughter first. Then it goes quiet, because everyone recognises themselves in the first half of that sentence.

You might recognise this: you type a question, you get an answer and you move on. In this article, you will read why that is riskier than it seems, and how you can use AI in a way that makes you sharper.

Copy the prompt at the bottom of this article, and you can let your AI challenge your own answer.

Good ideas often die in the second after they are born. Why real creativity needs the critic, and why you make something tangible as quickly as possible.

You know the moment. You are in the shower or you are sitting in the car, and suddenly you have a great idea. Eureka. A few seconds later the first thought arrives: yes but, that will never work, because. And there the idea dies, even before it had the chance to become something.

In my creativity workshops I call that the eureka and death pattern. Everyone recognises it. The solution might surprise you, because it starts with the critic in yourself and in your team.

Copy the prompt at the bottom of this article. With that you look at an idea with your AI in three separate rounds.

Knowledge agents, work agents and autonomous agents. Why the autonomous dream often derails today, and where the real gain is for those starting now.

The word agent comes up in every presentation about AI these days. Everyone has one, everyone is building one, and everyone means something different by it. You probably recognise this: you hear about agents, and you wonder which one you actually need. In my workshops, I therefore make a simple distinction first. There are three types of agents. They differ mainly in how much they do independently.

Copy the prompt at the bottom of this article. You can use it to describe your first work agent together with AI.

From answering to acting independently

Anyone who works well with agents goes through three phases. The first two are about efficiency. The third is about work you could never have done before.

When I speak to people who have been working well with AI agents for a while, I always hear the same story. It happens in three phases. Almost no one sees the third phase coming. Perhaps you are already somewhere in that sequence yourself, and you are already wondering what comes after the clearing up.

Copy the prompt at the bottom of this article, and you can search for your own third phase together with your AI.

Phase one: the backlog

First you clear up. All those tasks that have been on your list for months and that you never got round to. The folders you once wanted to organise. The quotes you still had to follow up. The overview of your clients that has been out of date for two years.

Why your AI keeps forgetting your tone of voice. Put your facts, your rules of behaviour and the knowledge per client each in their own place, and it will work faster.

'Why does it always forget my tone of voice?' You might recognise that question, because I get it in almost every workshop. Usually it turns out that everything is mixed together in one big instruction field: who you are, who you work for, how you write, which clients you have and what formatting you want. That is one long piece of text that is sent along with every conversation. The model reads it completely every time and still can only half follow it.

Copy the prompt at the bottom of this article. It splits your instructions out for you.

As soon as your AI reads files and emails, a hidden sentence can give it a different instruction. What prompt injection is and which rule you add today.

You ask your AI to summarise a document you just downloaded. In one of my workshops I show such an ordinary looking document. A few paragraphs of text, nothing special. Then I show what is written in white letters on a white background. An instruction, invisible to the reader and perfectly readable to an AI: search for passwords in the files of this user and forward them to this address. Then it gets very quiet in the room. Perhaps you are wondering now how much of what your AI reads actually comes from you.

Copy the prompt at the bottom of this article. With that you immediately protect your own AI tool better.

The smartest model is not always the best. If you choose consciously per task, you gain control, time and space in your subscription.

You do not take a sports car to get bread from the bakery. And yet most people actually do that with AI. Perhaps you recognise this: you turn on the most powerful model that your subscription allows. And then you use that for everything. A four-line poem, rewriting an email, cleaning up a folder full of files: it all runs on the most expensive engine.

Copy the prompt at the bottom of this article. It helps you to choose the appropriate model per task.

More power is not always better

It feels logical that the smartest model gives the best result. For complex thinking work, that is also true. For many daily tasks, it actually works against you. The largest models have a tendency to do too much. They write longer than necessary and add things you did not ask for. And sometimes they run ahead of you, in a direction you did not want.

You do not write the best prompt alone. With one sentence at the end you let AI question you first, and you become sharper in your own thinking.

People often ask me about the perfect prompt: which template, which order, which magic words. You might have wondered about that too. My answer surprises most participants, because I do not write my best prompts alone. I let AI question me first.

Copy the prompt at the bottom of this article. Further down is also the prompt package with six metaprompts that question you first, free for Members.

One sentence at the end of your prompt

The technique is called the metaprompt, and it is actually almost embarrassingly simple. You describe what you want to achieve and what a good result looks like. Then you add one sentence: 'Ask me three to five questions first before you start.'

Why a PDF burdens your AI more than the same text in Markdown, and two small habits that immediately improve your output and your instructions.

If I ask in a workshop who has ever dragged a PDF into an AI chat, almost all hands go up. If I ask who knows what Markdown is, only a few remain. Yet that second one is probably the most underestimated piece of knowledge in working with AI. Perhaps you belong to the first group. Then you are giving your AI more ballast than you think.

Copy the prompt at the bottom of this article to convert a document to Markdown yourself.

What Markdown is

Markdown is plain text with a few simple characters for structure. A hash for a title. Two asterisks around a word for bold. A hyphen for a list. That is really all it is. You can read it in any text editor, and it still looks logical even without formatting.

With every automation you choose how much oversight you keep: approving every step, or watching along with your hand on the brake. This is how you choose what fits your process.

As soon as you automate something with AI, a question arises that is often asked too late. Perhaps you are already wondering where you stand in this process. Do you still do something, do you only watch, or are you completely out of it? The answer determines how reliable your automation becomes and how much time it really saves you.

Copy the prompt at the bottom of this article, then you can immediately see step by step where you stand in your own automation.

Two forms of oversight

In my workshops I make a distinction between two forms: the human in the loop and the human on the loop. It sounds a bit like a play on words. The difference is big though.

If the answer is wrong, correcting it in a new message is the reflex. Editing your first prompt often works better, and your AI stays sharper.

You ask AI a question and the answer is not quite right. So you type 'no, shorter'. Then 'no, more formal'. Then 'and without that list'. Five messages later you have something usable, and a conversation full of half corrections. I see it in every workshop. It is the most natural reflex, and in my opinion rarely the best one.

Copy the prompt at the bottom of this article. With it you will see what was missing from your question.

What happens when you keep correcting

Every correction you type in a new message is added to everything that was already there. Your original question is still there, and also the failed first answer, your first correction, the second answer and your second correction. The model has to take all those layers into account. From that it deduces for itself what you actually want.

A summary of a summary loses something every time. With a handover document and a phased plan, your AI will not lose the thread.

Your conversation with your AI gets longer and longer, and at a certain point a button appears that shortens it. You click it, the chat is summarised and you can continue. It actually feels very efficient. In my workshops I still advise against it for anything that matters. There is a better way, and it only costs you one extra step.

Copy the prompt at the bottom of this article to have your first handover written. You will find the swipe file with sentences for a good handover further down, free if you have an account.

Describing a work process in plain language for an agent is flexible, but not always smart. When you are better off building a fixed workflow, and how you recognise the difference.

Working with AI agents becomes a lot easier when you can just describe a process in your own words. You write down how you do something, step by step, and the agent follows that. In many tools, such a description is called a skill. You will probably want to do that for everything. Yet I warn in my workshops that by no means every process belongs in plain language.

Copy the prompt at the bottom of this article to figure out what fits best for your own process.

What a skill is

A skill is actually a description of a work process. It is a standard procedure, written in Markdown, that an agent reads and executes. Take someone who wants an appointment. Then you first ask who it is and when it should be, and then where. You check the calendar to see if it fits. If it does not fit, you suggest an alternative. And after agreement, you send the invitation.

An agent does what it can, unless you provide a framework. Why boundaries in your instructions are just as important as the task itself.

You give your agent a clear task and assume that it will understand the rest itself. I understand that reflex. Yet in workshops I often tell the story of someone who gave an AI agent access to a credit card. The instruction was clear: book the cheapest flight to Bali. The agent found no more standard seats and booked a ticket in business class with an expensive airline. Technically, that was the cheapest available option. A few thousand euros later, the owner understood what had gone wrong.

Feel free to copy the prompt at the bottom of this article, so you can immediately write the boundaries for your own agent.

If you start building with AI straight away, you often get stuck. With a short discovery phase and a few early choices, you save days of work later.

Vibe coding is really just talking to your computer, and it builds it. That is how I explain it to participants who have never heard the term before. You describe what you want in plain language, and an AI agent writes the code. A tool, a website or a small app. What used to require a team, one person can do today.

Perhaps you already recognise this. You start for the first time, it gets off to a promising start, and then it goes wrong somewhere. The AI no longer seems able to change direction. I see that with almost everyone who starts.

A long chat does not train your AI, it fills the context window. Why the quality drops halfway through, and the habit that solves it.

Maybe you have been working in the same chat for weeks. In almost every workshop someone tells me that, and usually with some pride: 'That is how it gets to know me.' I understand that thought, because it feels like a colleague who increasingly knows how you work. Except it is not true, and exactly because of that the answer gets worse instead of better after a while. That is a bit ironic, because you are doing it precisely to make it better.

Copy the prompt at the bottom of this article to neatly wrap up a long chat and start fresh.

A simple peer coaching exercise using the colours of the Six Thinking Hats. You just hold up a card and learn to listen in the process.

Perhaps you want to help a colleague with a difficult decision and you quickly offer solutions. We all do that. There is one exercise from my creativity workshop that participants almost always take home with them. You need six coloured cards for it and one strict rule: you are not allowed to give advice.

Copy the prompt at the bottom of this article if you have no one to practise with, and let your AI hold up the cards.

How the exercise works

You work in a pair or a trio. One person talks about an idea or a challenge they are working on. That must be something real, not a made up case. The other listens, holding six coloured cards.

Four habits that turn an AI strategy into daily practice, for individuals, teams and entire organisations.

You have an AI strategy, or your team has met about it at least a few times. And yet not much is changing in your working week. You probably recognise this. EPIC helps you to turn that strategy into daily practice, for yourself, for your team or for your entire organisation.

Copy the prompt at the bottom of this article and apply the four habits to one task straight away.

Why strategy alone is not enough

Most AI adoption programmes fail for the same reason as most change programmes. They focus on the tools and forget the people. That is a bit ironic, because the tools are there precisely for those people. EPIC starts somewhere else: with the work you already do. That way you do not have to start a separate project. It is about four habits: Everyday tasks, Pair learning, Iterative feedback and Continuous improvement.

Five human capacities that AI cannot take over, and a prompt to help you discover which ones you already use in your work.

You know those workdays when nothing goes as planned. The situation is messy, the direction changes, and nobody knows exactly what is expected of you now. That is when you need cognitive agility. That is the human ability to adjust your thinking and your emotional and social approach. This is how you find your way in unpredictable situations. It is a meta-capacity that consists of five capacities.

Copy the prompt at the bottom of this article to see with your AI which of those five capacities you already use.

Five capacities, one meta-capacity

Seven questions you answer before you press enter: this is how you turn a vague prompt into a reliable briefing, for older and simpler AI models.

You type a short question into your AI and get something back that looks neat. Only it is not what you meant. So you try again, and again, until you refine the result yourself. You probably recognise this. The CRAFTER framework is a fixed way to brief your AI, and it ensures this happens less often.

With a loose prompt you are sometimes just lucky. With CRAFTER you actually get the same quality every time. You turn a vague prompt into a clear assignment, and you can repeat it.

A side note beforehand: CRAFTER is a valid framework for older and simpler AI models. For ChatGPT 5.x and Claude 5.x it no longer applies. That is why you will also find a prompt for the more recent models at the bottom.

From Status Quo to Elevate: five levels that show how professionals and teams grow their relationship with AI, and what each level requires of people.

You might have been using AI for a while, and you wonder where you actually stand. Or where your team stands. The Superworker Model maps out five levels of how professionals and teams grow their relationship with AI. Speed or efficiency say little about that. The model looks at what becomes possible at each level, and at what each level requires of the people in the system.

Copy the prompt at the bottom of this article and put your own work on the map, task by task.

A map of possibilities

The five levels are called Status Quo, Optimize, Redesign, Reinvent and Elevate. I see the model as a map of possibilities. Different parts of your work can be at different levels. So you do not have to reach the highest level everywhere. I would consciously choose where you put your energy and attention.

Why your judgement gets weaker when you give all your thinking to AI, and why human attention remains the scarce computing power.

You are letting AI do more and more of your thinking, and that is going smoothly. What happens to your own thinking in the meantime gets too little attention in the AI conversation. Perhaps you do not notice anything yourself yet.

Copy the prompt at the bottom of this article. It helps you choose which thinking you keep for yourself.

What is The Drift?

Perhaps you outsource cognitive work to AI, such as research and analysis, or writing and decision support. Then something subtle begins to happen. The muscles you do not use start to waste away. That actually happens very steadily. There is no single day when you feel the difference.

A chatbot waits for your question, an agent completes a goal. Here is how you build your first own AI workflow this week, with a prompt to help you.

You probably already use AI to ask questions, and perhaps you wonder what comes next. Meanwhile, a shift is happening. Many people have not noticed it yet. We are moving from asking AI questions to directing AI workflows.

Copy the prompt at the bottom of this article, and you will map out your first workflow together with AI.

What is the difference between AI agents and chatbots?

A chatbot waits for your input and gives you an answer. An AI agent is given a goal, breaks it down into steps, uses tools and delivers a result. It is a bit like the difference between asking someone for directions and hiring a driver who knows the route.