How do you use AI in your business?
You use AI in your business by picking one recurring task and doing it together with an AI assistant or AI agent. Five use cases apply to almost every company: answering customer questions, drafting texts, summarising meetings, analysing figures and making internal knowledge searchable.
For each use case below you will find what the AI does, which data you need and where you or a colleague checks the result. That shows you what each task asks of you before you start.
How many Dutch companies use AI?
In 2025, 17% of Dutch companies with two or more employed persons used AI, up from 8% in 2023. This comes from figures by Statistics Netherlands (CBS) published in December 2025. The 2025 figure is provisional.
Among small companies the share is lower: 14% of companies with 2 to 9 employed persons and 27% of companies with 10 to 49 employed persons used AI in 2025.
Of the companies that use AI, 35% do so for marketing or sales and 32% for business administration or management tasks, such as translation and planning. The five use cases in this article largely fall into those two groups.
1. Answering customer questions
Many customer questions cover the same topics, such as opening hours, delivery times, returns and invoices. An AI can draft an answer to those questions in your tone.
What the AI does: read the question, look up the answer in your information and write a reply. With an AI assistant you paste the question into the conversation yourself. An AI agent reads the message itself in the channel where it arrives.
What you need: a list of your most frequent questions with the answer you would give, plus your terms and current product information.
Where you check: at the start, let the AI produce draft answers that a colleague reads and sends. Questions about complaints, refunds and exceptions always go to a person. More on this use case is on the page about AI customer service.
2. Drafting texts and content
Newsletters, product descriptions, social media posts and proposal texts are work where AI delivers a first version that you finish.
What the AI does: write a first version based on your briefing, suggest variants and shorten or rewrite an existing text for another channel.
What you need: a briefing with target audience and tone, and a few examples of texts you consider good. The sharper the briefing, the less you rewrite afterwards. What is a prompt explains how to build such an instruction.
Where you check: facts, prices and promises to customers. A language model can make up a claim that sounds right. Read every text before it goes out.
3. Summarising meetings
After a meeting you want to know what was decided and who does what. AI turns notes or a transcript into a summary with action items.
What the AI does: list the decisions, name the action items per person and write a draft follow-up email.
What you need: your own notes or a transcript of the conversation. If you record a conversation, ask the participants for permission beforehand.
Where you check: whether the decisions are right and the action items sit with the right person. Read the summary before you circulate it, because a wrongly noted agreement carries through into what follows.
4. Analysing figures and reports
Sales figures, website statistics and customer feedback are available at most companies, while the time to go through them is missing. AI helps you get through an export faster.
What the AI does: read a spreadsheet or export, point out notable rises and falls and answer your questions about the figures in plain language.
What you need: an export with clear column names and a concrete question, for example which product group fell the most this quarter.
Where you check: recalculate the main outcomes in your own file. Remove personal data from the export if you use a tool your organisation has no agreements with.
5. Making internal knowledge searchable
Work instructions, manuals and agreements are often spread across documents, mailboxes and the heads of colleagues. AI makes that knowledge searchable with a plain question.
What the AI does: answer questions from colleagues based on your own documents and state which document the answer comes from.
What you need: current documents in one place. Outdated instructions produce outdated answers, so clean up first.
Where you check: appoint one person per topic who keeps the documents up to date. For questions about HR matters and contracts, the source document remains leading.
Which use case do you start with?
Start with the task that returns every week, for which you already have the data and where a mistake does little damage. For many companies that is drafting texts or summarising meetings, because you read every result yourself.
Work with that one task for a few weeks and note what goes well and where you need to adjust. Then pick the next one. If you have never worked with AI, start with the explanation in AI for beginners.
For most of these tasks an AI assistant such as ChatGPT or Claude is enough. An AI agent becomes interesting when a task returns daily and the AI needs to read or write in your systems itself. The difference is explained in AI agent vs ChatGPT.
Which data may you enter into an AI tool?
Only enter data into an AI tool that your organisation has made agreements about. The Dutch Data Protection Authority (Autoriteit Persoonsgegevens) reported in August 2024 that it had received several data breach notifications because employees had entered personal data of customers or patients into an AI assistant.
The regulator advises agreeing with employees which tools are allowed and which data they may enter. How to arrange that is covered in AI and privacy and in the article on AI and cybersecurity.
Frequently asked questions
What is the easiest way to use AI in your business? Pick one recurring writing task, such as a newsletter or a follow-up email, and let an AI assistant produce the first version. You read and improve the text. That way you learn to work with AI without anything reaching a customer unseen.
How many companies in the Netherlands use AI? According to Statistics Netherlands (CBS), 17% of companies with two or more employed persons used AI in 2025. In 2023 that was 8%. The 2025 figure is provisional and comes from a CBS publication of December 2025.
Do I need an AI agent or is ChatGPT enough? For tasks you start and check yourself, such as writing and summarising, an AI assistant such as ChatGPT or Claude is enough. An AI agent fits tasks that return daily and where the AI needs to work in your systems itself.
Which data do I need to start with AI? For most use cases you need examples of your own work: frequent questions with answers, texts in your tone, notes or an export with figures. Only enter personal data if your organisation has agreements about it.
Does a person always need to check AI output? Yes, for everything that goes to customers or that a decision follows from. A language model can give a wrong answer that sounds convincing. Check facts, amounts and agreements before you use the result.
Deploying an AI agent yourself?
Browse our ready-to-go AI agents or have a custom agent built on your own infrastructure. Want to know the cost first? Read the AI agent pricing guide or book a free intro call.
