When artificial intelligence is discussed, headlines about tech giants and billion-dollar investments dominate. The impression quickly forms: AI is something for Google and SAP, not for the skilled trades business in Flensburg or the freight company in Potsdam.
That is not true.
In the German SME sector, thousands of companies are working productively with AI today. Not as a pilot project, not as innovation theater, but as part of their day-to-day business. And the solutions are often surprisingly pragmatic: no in-house data science team, no million-euro budgets. Instead: a concrete problem, an existing tool, a measurable result.
Five examples show what this looks like in practice.
1. Skilled trades: writing quotes in 10 minutes instead of 45
An electrical business with 25 employees near Lübeck knows the problem: the master electrician drives out to the customer, takes photos, notes the requirements, and then spends another hour in the evening preparing the quote. With three to four appointments a day, things pile up. Customers wait, orders are lost.
For several months now, the business has been using ChatGPT via the business API with a data processing agreement, combined with a simple template. The master electrician does not enter customer names or addresses into the prompt; those are only added to the finished quote. After the appointment, he dictates his assessment into his phone, and the system turns it into a structured quote draft with line items and price calculations based on the company's own price lists. Review, adjust, send.
The result: the average time per quote dropped from 45 to around 10 minutes. 30 % more quotes go out per week. Quality is consistent because the system draws on the business's historical calculation data. And the master electrician has his evenings back.
What made the difference was not the technology, but the integration into the existing workflow. No new portal, no complicated interface. Voice input, done.
2. Services: customer service with an AI chatbot
An insurance broker with 40 employees in Kiel handles hundreds of customer inquiries every day: status queries on current policies, questions about coverage amounts, change requests, claims reports. The phone never stopped ringing, emails piled up, and the average response time was four hours.
The company implemented an AI chatbot on its website and in its customer portal, hosted on European servers with GDPR-compliant data processing. The bot was trained on the most common question categories and connected to the existing CRM system. It reads personal data from policy records out of the CRM without passing it on to external AI services. It answers standard questions immediately and forwards complex matters, with all relevant information, to the right case handler.
The result: 65 % of all inquiries are handled fully automatically. The average response time dropped from four hours to three minutes. Customer satisfaction rose measurably because simple matters are resolved immediately. And the staff finally have time for the cases that truly need personal advice.
Important: the chatbot does not replace personal contact. It filters out the routine so that the personal advice gets better.
3. Accounting: processing invoices automatically
A wholesaler for catering supplies in Brandenburg an der Havel, 30 employees, receives 40 to 60 incoming invoices every day. Previously, a clerk entered each one manually: invoice number, amount, supplier, cost center. Half a day went into that alone.
Today, the invoices arrive by email and are automatically read by an AI-based document processing tool that runs on the company's own servers in Germany. The system recognizes the relevant fields, assigns the invoice to the right supplier and the appropriate cost center, and creates it in the accounting system. The data never leaves the company. The clerk now only reviews exceptions and special cases.
The result: manual entry time fell by 80 %. Error rate in assignment: under 2 %. The clerk uses the freed-up time for dunning and liquidity planning.
None of these examples requires a data science team or a custom-built solution. Most AI solutions in the SME sector are based on existing tools and platforms, configured correctly and integrated into everyday work.
4. Sales: emails and letters at the push of a button
An IT systems house in Rostock with 50 employees had a typical sales problem: the field sales staff spent hours every week writing follow-up emails, quote cover letters, and customer correspondence. Everyone wrote in their own style, some better, some worse. And it usually took too long.
The team now uses an internal GPT-based tool via the API (with a data processing agreement, and no model training by the provider) tuned to the company's communication style and product range. The sales reps enter bullet points (context of the conversation, next step); customer names and personal details are only added to the finished draft. The system generates a draft in the right tone. Adjust, send.
The result: the average time per customer email dropped from 20 to 5 minutes. The follow-up rate rose by 40 %, because the sales reps now actually send the emails promptly instead of putting them off. And the quality is more consistent, because everyone works from the same foundation.
5. Accounting: capturing and assigning receipts automatically
A tax advisory firm in Eckernförde with 15 employees processed around 2,000 incoming receipts per month for its clients. Each invoice had to be opened manually, the amount read off, the cost center assigned, and the entry transferred into the accounting system. Three to five minutes per receipt. That added up to several working days per month.
The firm now uses a tool with integrated AI text recognition (in this case DATEV with receipt image recognition and assignment suggestions). Receipts are scanned or uploaded as PDFs. The AI automatically reads the invoice number, amount, date, and sender, suggests the appropriate booking category, and assigns the receipt to the right client. The employee briefly reviews and confirms.
The result: receipt processing now takes under a minute apiece on average. The team saves two full working days per month. The error rate in assignment has dropped, because the AI works more consistently than manual entry after the third coffee. And the staff use the time gained for client advisory work instead of data entry.
Conclusion: concrete problem, pragmatic solution, measurable result
Five companies, five industries, five completely different use cases. What connects them: all are running productively. All deliver measurable results. And all follow the same pattern.
First: there was a concrete, clearly defined problem. Not "we want to do something with AI", but "our quote preparation takes too long" or "our invoice entry ties up too many staff".
Second: the solution was pragmatic. No moonshot project, no two-year development cycle. Existing tools, configured correctly and integrated into everyday work.
Third: the result was measurable from the start. Fewer minutes per quote, fewer errors in invoice entry, faster response times, better margins.
And perhaps most importantly: in none of these cases did AI replace people. It relieved them. The master electrician writes quotes faster, the clerk focuses on exceptions, the sales rep actually sends his emails. AI in the SME sector doesn't work despite the people; it works because of them.
If you are wondering whether AI has a concrete lever in your company too: the answer is almost always yes. The only question is where the biggest effect lies.
