AI in practice
ChatGPT for Business Owners: My Use Cases from Daily Practice
A while ago, a client told me: “The AI doesn’t sound like me.” I asked him: “Did you tell it how you sound?” Silence. He had given ChatGPT exactly one sentence: “Write me a good social media post about meditation.” No example, no goal, no style. The result was generic.
That is exactly how most business owners use ChatGPT. Like a search engine. One question, one answer, close the tab. But inside the tool sits a digital employee. In this article, I show you how I use AI in business. With concrete prompts, real examples and a clear order.
One thing first. I work with several AI tools. Each has its own strengths, its own style, its own character. The principles in this article apply to all of them. They work with ChatGPT just as well as with Claude or Gemini. You can read which tools I use for what in my article on AI tools for solopreneurs.
Why most business owners use ChatGPT the wrong way
I know the IPO principle from my time in IT: input, processing, output. It explains almost every bad AI result.
“Write me a social media post,” “Write me a social media plan” and “Write me a social media strategy” are three different inputs. You get three different results. The processing is the model you choose. That changes the output too. If you don’t like the result, it is almost always because of your input or the wrong model.
The lever is context. For me, a good prompt has four parts:
- Instruction: What exactly ChatGPT should do.
- Goal: What you need the result for.
- Positive examples: This is how it should look.
- Negative examples: Not like this, please.
For the meditation client above, the solution looked like this:
Here are three texts I wrote myself: [insert texts].
Analyze my style: sentence length, tone, typical phrases.
Here are two AI texts I don't like: [insert texts].
Briefly explain the difference.
Then write a post about [topic] in my style.
Goal: comments and messages from potential clients.
It is better to give too much information than too little. ChatGPT can’t read your mind. It can only work with what you give it.
Choosing the right model, as of August 2026
Since July 2026, ChatGPT runs on GPT-5.6. There are three variants. Sol is the flagship and thinks deeply. Terra is the middle ground for daily tasks. Luna is fast and cheap. Since August 6, 2026, it has been the default in the free version.
Here are the prices as of August 2026. The free version costs nothing. Go costs 8 dollars, Plus 20 dollars. Since April 2026, Pro comes in two tiers, starting at 100 dollars a month. The Business plan for teams is 25 dollars per user, or 20 dollars with annual billing.
My rule: for anything that makes money or matters strategically, use the smart model, which is Sol. Many people run the default model out of habit and then wonder why the answers are shallow. That is like letting the intern develop your pricing strategy. For quick routine tasks, Luna is all you need.
My use cases: ChatGPT for business owners in practice
Marketing: texts and images that sound like you
Marketing is where business owners see results fastest. Ad copy, emails, landing pages, social media posts. The process is always the same. First set up your style and target audience as context, then produce, then give feedback.
For this, I create a project in ChatGPT. There I store instructions, my target audience, my offer and sample texts. Every new chat in this project knows this context. I don’t have to start from scratch each time. I also create images for posts and ads right in the tool.
In my article on how to write sales copy, I show you how to write texts that sell instead of just sounding nice.
Sales: qualify inquiries instead of chasing everyone
Here is a use case from my weekly sessions with business owners. Inquiries come in by email, and nobody knows which ones are any good. The solution is a scoring prompt with fixed criteria.
Here is a customer inquiry: [insert email].
Rate it on a scale from 0 to 10 using these criteria:
budget is visible, problem fits my offer,
timeline is realistic, the decision-maker writes personally.
Explain the rating in two sentences.
From 7 points up: write a personal reply with a suggested meeting time.
This way, you first work on the inquiries most likely to bring in revenue. You’ll find more strategies in my article on how to get customers with AI.
Content: turn existing material into new products
A business owner I know had run workshops for years. Dozens of PowerPoint presentations, carefully worked out, but unused. Now they are turning into a book. The AI reads all the files, structures the material and suggests an outline.
The most important step: tell the AI what you need the result for. Do you want to make money with the book, position yourself as an expert or win leads? The more precise the goal, the better the help. Here is a trick I always use:
Rate each document on a scale from 0 to 10.
Criterion: How valuable is the content for [target audience and goal]?
List the top 20 and give a short reason for each.
This way, the AI separates your best material from the rest before you write a single line.
Bundle your knowledge: your second brain
I had about 6,000 notes scattered across different apps and folders. Ideas, concepts, half-finished articles. Many business owners are in the same spot. Their knowledge sits in Evernote, Dropbox and old folders. ChatGPT helps you organize this knowledge and put it to use.
In practice, this works through Projects and app connections. For example, you connect Google Drive, upload files and then ask questions about your own material. For in-depth research, I use Deep Research. The feature researches the web in several steps, reads sources and delivers a report with citations. It saves me many hours of market research, for example on competitors or on the language my target audience uses.
Admin: invoices, reports, offers
The least exciting use case often saves the most time. Pre-sorting invoices, summarizing bank statements, structuring offers, boiling long documents down to the key points. Tasks that come up every week and need no creativity.
A business owner recently asked me if he is even allowed to use AI for this, because of company data. You’ll find my answer below in the section on data protection. The short version: stay pragmatic, anonymize, use the Business plan.
ChatGPT as a thinking partner for strategy
This is the use case almost nobody uses. The new models no longer just answer questions. They question them. Here are three prompts I use regularly:
What questions should I ask you that I have never asked you before?
Where do you see blind spots in my business
that I don't see myself?
What is the biggest financial domino I could knock over now
so that all other steps become easier?
When I first asked one of the current top models these questions, I was stunned. The answers went beyond my question. The model challenged my assumptions, like a smart sparring partner. Here, too, context is the requirement. The more the model knows about your business, the better the sparring.
Lead ChatGPT like an employee
I used to be an IT project manager. I ran an international campaign in eight countries, with 20 to 40 people at the same time. Every morning at 9 a.m., I sat in a meeting. Everyone wanted feedback. On the design, on the platform, on the commercial. When I put off feedback, the project stood still.
With ChatGPT, it is exactly the same. The AI delivers fast. The bottleneck is you. Three things make the difference:
- Communicate clearly. Say precisely what you want. Never assume the AI will guess it. I often dictate my instructions. That is faster than typing and automatically gives more context.
- Give guidelines. Tell ChatGPT how it can recognize a good result. For example: “At the end, check that every paragraph contains a concrete action.” Otherwise, it will ask you at every step along the way.
- Use the feedback loop. If you don’t like a result, don’t throw it away. Give it back as a negative example and explain why it doesn’t fit. Every round gets better. Collect the good results as positive examples for next time.
If you can think clearly and say things clearly, you get many times more out of ChatGPT. For business owners, this skill matters more than any single tool.
Automate what already makes money first
The most common mistake I see: business owners start with exciting new ideas instead of their own core business. My basic rule is different. First automate the things that already make you money and eat up a lot of time.
The weekly report, writing offers, follow-up emails, preparing the bookkeeping. If you gain two to five hours a week there, invest that time in sales or new offers. Only then is it worth experimenting with new things.
By the way, for automations that go beyond the chat, I work in the terminal. You can find an overview of all my AI articles on the AI overview page.
Data protection: pragmatic, not panicked
Yes, you need to be careful with data protection. Three rules are enough for daily work. Don’t enter your customers’ personal data into the free version. Anonymize names, amounts and account numbers when you upload documents. And use the Business plan for company work. There, OpenAI does not use your content for training by default.
The other side matters just as much to me. I see business owners who do nothing at all because they fear data protection issues. If you reject every cloud service on principle, you would also have to question Gmail and your own invoicing software. Keep the risks in proportion, take the basic measures and solve problems when they really come up. Losing your competitive edge is a risk too.
Start with one task
ChatGPT for business owners does not mean rebuilding everything at once. Pick one task that comes up every week and annoys you. Give ChatGPT context, examples and a clear goal. Give feedback until the result is right. Then save the process as a project and take on the next task.
After a few weeks, you have a system instead of a toy. In my article on AI for solopreneurs, you can read how to build a complete setup for your business on top of this.