Finding AI use cases: how to spot where AI helps your business
5 min read
Short answer
You will find good AI use cases where your team regularly reads, sorts, summarises or writes texts. Ask the people who do this work every day, collect concrete tasks rather than general ideas, and check each one for volume, time spent and consequences of errors. This usually yields two or three candidates that can be tested quickly.
What does this mean for your company? Book an intro call
The best AI ideas rarely come from management or from a software vendor. They come from where the work gets done. Your team knows exactly which tasks take a lot of time and bring little satisfaction. What is usually missing is an eye for which of them suit AI.
How do you recognise potential for AI?
Look for these patterns in daily work:
- Reading to extract little: someone reads a multi-page document to type three details into a form.
- The same questions again and again: customers or colleagues regularly ask similar things, and the answer is somewhere in your documents.
- Texts built from blocks: quotes, reports or replies are created by copying and adapting.
- Sorting and forwarding: incoming items have to be read before it is clear who is responsible.
- Knowledge in a few heads: only two people know where something is, and both are in constant demand.
How do you gather ideas with your team?
Invite staff from two or three departments to a workshop of about ninety minutes. Good questions for it are:
- Which task costs you the most time every week?
- What would you most like never to do again?
- Where do you regularly wait for information?
- Which questions do you answer over and over?
Then walk through a typical week. Concrete tasks such as “matching supplier confirmations with the purchase order” are far more valuable than general ideas such as “AI in purchasing”.
How do you assess the ideas?
Four questions count for each task: how often does it occur? How long does it take? What happens if there is an error? Is the necessary data available digitally? How to turn this into a ranking is shown in Which process should you automate first?
An honest check belongs here too: not every idea needs AI. If a process follows fixed rules and the data is clean, classic automation is cheaper and more reliable. AI plays to its strengths with language, documents and changing formats.
Examples from different departments
- Sales: extracting enquiries from emails, preparing draft quotes.
- Purchasing and accounting: reading and matching invoices and delivery notes.
- Customer service: sorting requests, drafting replies from the knowledge base.
- Engineering and quality: writing inspection reports, making manuals and standards searchable.
- HR: drafting job ads, answering questions about internal policies. Be careful with selecting applicants: under the EU AI Act that counts as a high-risk application.
The next step
Choose the most promising candidate and test it with real examples from your daily work. Which applications tend to pay off is covered in Which AI projects pay off?
Julius Launhardt facilitates workshops to find AI use cases and then assesses the ideas for benefit and effort. More on the AI consulting page.
Frequently asked questions
How long does finding use cases take?
For a first overview, a workshop of about ninety minutes and a few conversations with the departments are enough. Afterwards you usually have a handful of concrete candidates.
Do I need technical knowledge for this?
Not for collecting ideas. What matters is knowledge of your processes, and your team has it. Technical experience is only needed when assessing effort and feasibility.
When is classic automation better than AI?
When the data is already structured and the process follows fixed rules, for example transferring orders between two programs. A rule-based solution is then cheaper and more reliable.
What do I do with ideas that do not fit?
Write them down. Some ideas become interesting when a system is replaced or volumes grow. A look at the list every six months is enough.