Short answer: Only four percent of German trade businesses use AI. The fastest returns come from phone assistants for missed calls, automated receipt capture and quotation drafting. All three save office time rather than site time, and office time is where the bottleneck sits today.
The gap is wider than it needs to be
The skilled trades are not a small sector. The German Confederation of Skilled Crafts counts roughly 1,038,000 registered businesses, around six million employees and revenue of about EUR 783 billion for 2025. Even so, there is an unusually wide gap to the rest of the economy when it comes to AI.
The Bitkom study on digitalisation in the skilled trades surveyed 504 trade businesses with at least one employee by telephone (surveyed in 2025, published January 2026). The result: 4 percent use AI, 9 percent plan to, and 84 percent say it is simply not a topic for them. For comparison, the DIHK digitalisation survey 2026 covering 4,686 companies across the economy found 34 percent already using AI and another 35 percent planning to.
The reasons sit in the same study. 72 percent of businesses say they have too much on to deal with digitalisation. 58 percent have no overview of what is even possible. And only 29 percent have staff who can work with AI. That is not indifference, it is a capacity problem.
Where the time actually goes
The answer is almost never on site. It is in the office and on the phone. A survey by the Erfurt Chamber of Skilled Crafts in its own district found that owner-managers spend an average of 33 percent of their working time on activities driven by bureaucratic requirements, rising to 43 percent in larger businesses. The figure comes from a regional survey and is not nationally representative, but it matches what owners describe in conversation.
Add the staffing shortage. The ZDH estimates around 250,000 vacancies in the skilled trades, plus roughly 20,000 unfilled apprenticeship places on average. If you cannot hire, you can only redistribute work or eliminate it. AI is no substitute for a qualified tradesperson, but it can get the owner away from the desk.
The four use cases that pay back first
| Use case | Setup effort | Typical running cost | Effect |
|---|---|---|---|
| AI phone assistant for missed calls | 1 to 2 days | from around EUR 60 to 80 per month | no enquiry gets lost |
| Receipt capture with automatic recognition | 0.5 days | included in accounting packages, from around EUR 8 per month | less re-keying, clean pre-coding |
| Quotation drafts from measurements and text blocks | 2 to 4 days | included in trade software | quotes go out the same day |
| Site documentation by voice and photo | 1 day | included in trade software | records without evening shifts |
1. The phone assistant
The single biggest lever in the trades is not really an AI topic, it is an availability topic. If nobody picks up, the caller rings the next business. Several trade software vendors have responded by building in voice assistants: HERO Software includes HERO Voice from its mid tier with 60 free minutes a month, and plancraft offers its AI reception with a base allowance in every package and as a separate plan beyond that.
Expectations matter here. An assistant of this kind does not replace a consultation. It captures the name, the callback number and the request, files it correctly and stores it in the system. That is enough to save the call.
2. Receipt capture
Accounting tools such as Lexware Office do not market themselves with the word AI, but they do exactly this: photograph the receipt, recognise it, propose the booking, reconcile the payment automatically. For a business handling two kilos of paperwork a month, this is the least glamorous and most reliable time saving in the whole field.
3. The quotation draft
This is where the revenue lever sits. A quote that arrives three days later is often already lost. With measurement data, a schedule of services and text blocks, a solid draft can be generated in minutes, leaving the owner to check it and adjust the pricing.
4. Site documentation
Dictation instead of a form. Voice notes and photos are turned automatically into a structured site diary or job report. The value only shows up in the first dispute about variations, but then it shows up clearly.
Worked example: what missed calls cost
A business with twelve employees misses four calls on an ordinary working day, because everyone is on site or the office line is busy. Across 220 working days that is 880 unanswered calls a year.
Let us be deliberately conservative. Only 10 percent of those are genuine new enquiries, so 88. Of those, 25 percent turn into a job, so 22 jobs. At an average order value of EUR 900 that is roughly EUR 19,800 in annual revenue that simply goes to a competitor.
A phone assistant at EUR 80 a month costs EUR 960 a year. The break-even sits at 960 divided by EUR 225 of expected value per recovered enquiry, so about 4.3 recovered enquiries a year. Even if the assistant rescues only a tenth of the lost enquiries, it has paid for itself many times over.
The inputs are assumptions, but the structure of the calculation holds for almost any business. Put your own numbers in before you buy anything.
The deadline that is coming anyway: e-invoicing
Quite apart from AI, there is a reason to touch your office software now. According to the German Federal Ministry of Finance, domestic businesses have had to be able to receive e-invoices since 1 January 2025, explicitly with no exemptions. Even small businesses that are not required to issue e-invoices must be able to receive them; an email inbox is sufficient for receipt.
For issuing invoices, transitional rules apply: generally until the end of 2026, and until the end of 2027 where prior-year revenue was up to EUR 800,000. Since 2025 a plain PDF no longer counts as an e-invoice; structured formats such as ZUGFeRD or XRechnung are required.
If you are changing software anyway, check AI features and e-invoicing capability in one go rather than migrating twice.
How to start without turning it into a project
- One week of tally marks. Record missed calls, late quotes and evening hours in the office. Without that list you are deciding on gut feel.
- Pick one use case. The one with the most marks, not the technically most interesting one.
- Test for four weeks. Almost every vendor offers a trial. Use it with real jobs, not demo data.
- Then roll out. With one person who can explain it, otherwise it quietly stops.
If you are unsure where you stand, our AI check helps you place yourself. For a cross-sector order of play, see five processes small teams should automate right away.
Frequently asked questions
Do I need new trade software for this?
Not necessarily. A phone assistant and receipt capture can usually run alongside your existing software. Only for quotation drafting and documentation is a switch or upgrade worthwhile, because there the underlying data matters.
Will customers notice that an assistant is answering?
Yes, and they should. Since August 2026 the EU AI Act requires that people can tell when they are interacting with an AI system. In practice this is no disadvantage: most callers prefer an assistant that reliably arranges a callback over a ringing tone.
What about customer and staff data?
Check the data location and make sure a data processing agreement is in place. With voice recordings, retention periods come into it as well. Ask before you sign, not after.
Is this worth it with three employees?
For the phone assistant and receipt capture, yes, often especially so, because nobody is sitting in the office. Quotation automation needs a certain volume of similar quotes before the setup pays back.
Is there any support for getting started?
Germany’s Mittelstand-Digital centres provide AI trainers nationwide whose services are free of charge and vendor-neutral. For a first orientation without a budget, that is a good starting point.
Conclusion
Four percent adoption does not mean AI fails in the trades. It means hardly anyone has had time to try. Start small, at the phone or with the receipts, and you will see the benefit within weeks without changing anything about how the work on site is done. If you would like company on the way, look at our automation services and managed service, or get in touch. On phone and chat availability, our article on AI assistants in customer service is a useful companion. Sector figures: key figures of the German skilled crafts (ZDH) and ZDH on securing skilled labour.
