There is one sentence I hear more often than any other. It usually comes after the third or fourth attempt, and it almost always sounds apologetic:
“Kirsten, I have no idea what I should concretely do with AI.”
That is not a beginner’s sentence. It comes from women with a running business who have long tried things. They have watched videos, saved prompts, tested tools. The folder with the tutorials is full. Still, nothing has happened.
I believe that is not down to you. It is down to almost everyone answering the wrong question.
Why the usual answers do not help
If you ask what can be done with AI, you get a list. Ten tools, fifteen prompts, seven use cases. All correct, all well meant — and still you end up back where you were before.
The reason is simple: A list answers the question “What would all be possible?” But your actual question is “What do I do on Monday morning?”
Between those two questions lies the whole difference. The first produces possibilities — and possibilities can be collected. The second produces a decision, and decisions are uncomfortable because afterwards you have to do something.
That is why so many keep collecting. Not out of laziness, but because collecting feels like progress and demands no decision.
The question that helps instead
Turn it around. Do not ask what AI can do. Ask:
Which task do I do again and again — describing the same thing every time?
That is the point where something becomes handable-over. Not because the task is unimportant, but because it follows a fixed pattern: you gather a few pieces of information, apply your knowledge to them, and out comes a result in always the same form.
That is exactly what a machine can take over. Not your judgement — but the path from information to result.
An everyday example — deliberately without business
Let me show you with an everyday task everyone knows and nobody likes: finding a fitting gift.
If you think about how you do that, it always runs the same way. You consider who it is for. What occasion. How much you want to spend. What the person likes. And what is absolutely off the table — the candle she has already received three times.
Five inputs. Then you think and come up with a few ideas. That used to take me half an hour, often with a mediocre result.
I wrote that routine down and handed it over. What came out is Schenkgenie: you answer the same five questions and get ten suggestions — each with a rationale and a price range. Free, no sign-up.
Try it before you read on. Seriously. It takes two minutes, and the rest of this article makes more sense afterwards. If nobody comes to mind right now: there are ready-made pages for the typical cases, such as gift ideas for grandma.
What matters about this — and what does not
The example is deliberately mundane. It is not about gifts, and it is not about me having built something either.
It is about what happened between the task and the result: someone once cleanly wrote down which inputs are needed, what should happen with them and what the result must look like.
That is the whole work. Not the technology.
And because it has so little to do with technology, it works exactly the same for you — with the tasks you have been doing again and again for years.
What this looks like in your business
Three examples I find in almost every business. You will probably recognise at least one.
The offer. A prospect enquires. You look at what she needs, consider scope and price, and write an offer — new in content, but in form the same as always. That is a pattern: enquiry in, structured offer out.
The first call. You talk to a potential client and ask essentially the same questions. Afterwards you assess whether she fits and what she would need. That too is a pattern — and the follow-up work even more so.
The follow-up. Someone received an offer and has gone quiet. You write a reminder, friendly, not pushy, referring back to the conversation. For the fiftieth time, always newly phrased, always along the same pattern.
For all three: your judgement stays with you. What can be handed over is the path from the inputs to the finished text.
How to proceed
Four steps. No tool comparison, no preparation that takes weeks.
1. Pick one task — exactly one. Take the one you do at least twice this week and that annoys you most. Not the most important one. The most tiresome one.
2. Write down how you do it. As if you were explaining it to a new employee: which inputs do you need? What do you do with them? What does the result look like? How do you tell it is good? That is the part almost everyone skips — and the one that decides between success and failure.
3. Hand it over and check. Look at the first version, correct, sharpen. Do not expect it to land immediately. Expect it to land after the third round.
4. Actually use it the next time. That is the step where it usually fails. Do not start from scratch again — take what is already there.
If you like, you can file the result so it stays permanently retrievable — how that works, I have described here.
Why it still often does not happen
Three reasons, all understandable.
“I am not techie enough.” You need no technology. You need the ability to describe your own work — and you have that, otherwise you would not have a business. In the gift example, the hard part was finding the five right questions. Not the building.
“I am waiting until I have found the right tool.” The tool is rarely the problem. Start with the one you already pay for. What that costs and which access is enough, I have worked out here.
“I just do not have time for this right now.” The most honest objection — and the most expensive. The task you do twice a week costs you over a hundred repetitions a year. One afternoon now against a hundred repetitions later.
What I want to leave you with
You do not need an overview of AI to use AI. You will not get one anyway; the field changes faster than anyone can keep up reading.
What you need is one single task that runs differently next week than this week. One. After that you know from your own experience how it works — and the second one takes only half as long.
That is the difference between collecting and implementing. And it is smaller than it looks from the outside.
And if you would rather not do this alone: In my free community, solopreneurs show which task they handed over first — what worked and what did not. Often it is enough to see what someone else started with.
to the community →Frequently asked questions
What can I concretely do with AI as a self-employed person?
The most sensible start is a task you repeat regularly and that always follows the same pattern: writing offers, following up on first calls, answering enquiries, chasing up. Not the most important task, but the most tiresome one. What decides is not the tool, but that you once cleanly describe which inputs are needed and what the result should look like.
What all can you do with AI?
Honestly: more than any list can capture — and that is exactly why the question gets you nowhere. More useful is the inversion: which of your tasks follows a fixed pattern? Texts with an always-identical structure (offers, follow-up emails, summaries), evaluations along always-identical criteria, research with always-identical questions — those are the candidates where AI carries immediately.
How do I use AI sensibly in everyday life?
By the same principle as in business: on recurring tasks with a fixed pattern. The gift example above is exactly that — an everyday task with five always-identical inputs. Meal planning, trip preparation or pre-sorting documents also run along patterns. Sensible here means: not a little bit everywhere, but one task completely.
I am not techie — can I still do this?
Yes. The hard part is not the technology but describing your own work: which information do you need, what do you do with it, how do you recognise a good result. You can do that, otherwise your business would not be running. The AI takes care of the implementation.
What should I start with if I already have too many tools?
With no new one. Take the tool you are already paying for anyway and apply it to exactly one task. Another tool in this situation is not progress but an extension of the collecting.
How long until this pays off?
With a task you do twice a week, you reach over a hundred repetitions a year. If you invest one afternoon and afterwards save ten minutes per pass, you have the investment back within a few weeks. More important than the math, though, is that after the first task you know how it works — the second takes only half as long.
How do I recognise that a task is suitable for handing over?
By three traits: you do it regularly, you always need the same inputs for it, and the result always has the same form. If all three apply, it is suitable. As soon as every repetition looks completely different or your personal judgement is the core, it is better kept with you.
