AI for trade businesses: 7 checks before you choose

Which AI tools for trade businesses fit yours depends less on the product name than on seven questions: where the data is held, what the tool can read and write in your trade software (Handwerkersoftware), who maintains it, how the costs grow, how you get out again, what the EU AI Act (KI-Verordnung) requires of you, and who checks what the AI writes into a quote. You will not find a ranking here. We build AI solutions ourselves, so our offer is set out separately in its own section at the end.

AI in the trades: what businesses do with it today

Most uses concern the office, not the job site. In a survey by the trade software vendor plancraft, 48 % said they use AI for communication and 35 % for preparing quotes and invoices (Handwerksblatt, 14 Aug 2026). This is a vendor survey, put in context below in the section on adoption figures.

Typical examples are calls answered while everyone is on site, pre-drafted emails, quote texts written from bullet points, and job ads. Whether any of this pays off for you depends on your volume, not on the technology.

Examples: use cases from office to job site

ChatGPT and similar tools for trade businesses: what works and what doesn’t

For texts without customer data, such as a job ad, a social media post or the outline of a quote, a general chat assistant is often enough. Things change as soon as customer names, addresses, phone numbers or photos go in. The provider then processes personal data on your behalf, and you need an account covered by an AVV (data processing agreement, Art. 28 GDPR) (as of 20 October 2026). Whether your plan covers this is stated in the provider’s contract documents. Without an AVV, no customer data goes in, not even into a free account.

A chat window is also not connected to your trade software. Whatever it produces, you copy across by hand. We use OpenAI, the provider of ChatGPT, ourselves (privacy policy, section 17). What to clarify before you start is summarised in the GDPR checklist for AI in your business.

Seven checks

1. Data location and AVV

Ask where data is stored and processed, which subprocessors are involved, whether an AVV is in place, how long recordings and transcripts are kept and how they are deleted. How to check providers on this is covered in GDPR-compliant AI providers for SMEs.

2. Interface to your trade software

“Has an interface” does not yet mean “can book appointments”. Ask to be shown which data the tool can read from your trade software and which it can write: create contacts, enter appointments, attach notes to a job. Also clarify who applies for access with your software vendor and whether it costs anything.

3. Effort and maintenance

Prices, services, opening hours and holiday cover change, and an AI tool only knows about it if someone updates it. Ask who does that, how quickly and at what cost. We found no figures on maintenance effort in hours. On the starting point in the team: in the Bitkom study of the trades, 29 % of businesses said they have staff who can handle AI. Among companies with 20 or more employees (not a trades figure), 66 % rate their employees’ AI skills as low (Bitkom, 14 Sep 2026).

4. Usage-based costs

Ask whether billing is by minutes, messages or consumption, whether there is a cap and what happens in a seasonal peak. The licence price is rarely the full amount: among AI users with 20 or more employees, 51 % name infrastructure such as cloud and computing power as a major cost item, 50 % data preparation and 41 % integration into existing systems, while only 21 % name subscriptions and licences (Bitkom, 14 Sep 2026, not a trades figure). How the costs of a phone assistant break down is covered in What does an AI phone assistant cost?.

5. Lock-in and exit

Clarify before you sign: term and notice period, export of your data in a format you can reuse, handover of the call script and knowledge base, who owns the phone number and when data is deleted after the contract ends.

6. The AI Act for users

As of 20 October 2026. A business that uses a ready-made AI system is usually a “deployer” (Betreiber) under the AI Act, meaning someone who uses an AI system under their own authority (Art. 3(4)). Under Art. 3(3), the provider is whoever develops an AI system or has it developed and places it on the market or puts it into service under their own name or trademark, whether for payment or free of charge. Who is obliged depends on your role: if you have a system built for you and put it into service under your own name, you may be the provider yourself. Four points matter to you:

• Prohibited practices (Art. 5): The prohibitions also cover the use of an AI system, so they apply to deployers too. The obvious one for businesses is Art. 5(1)(f), which has applied since 2 February 2025: AI systems that infer the emotions of a person in the workplace or in education institutions are prohibited, unless the use is intended to be introduced for medical or safety reasons. Art. 5(1) lists further prohibitions; some of them apply only from 2 December 2026.
• AI literacy (Art. 4): Since the amendment by Regulation (EU) 2026/1744, in force since 27 July 2026, the text reads (our translation of the German version): providers and deployers “take measures to support the development of the AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf”. And further: “This obligation does not oblige providers or deployers to guarantee a particular level of AI literacy for any person.” A short, documented briefing is an obvious place to start.
• Telling people it is AI (Art. 50(1)): The duty falls on the provider, which, depending on your role, may be you. It must design AI systems that interact directly with people so that they are informed they are talking to an AI, unless this is obvious to a reasonably well-informed, observant and circumspect person from the circumstances. A limited exception applies to AI systems authorised by law for law enforcement. Either way, we recommend saying so in the greeting, because you decide what the greeting says.
• Deepfakes (Art. 50(4)): This duty falls on you as the deployer. If your AI system generates or manipulates image, audio or video content that resembles existing persons, objects, places or events and would falsely appear authentic, you must disclose that the content has been artificially generated or manipulated. For evidently artistic, creative, satirical or fictional works, a disclosure that does not hamper the display is enough; the duty does not apply where use is authorised by law for law enforcement. Art. 50(4) also contains a rule for AI-generated text published to inform the public on matters of public interest. Whether an avatar video with your face and voice counts as a deepfake in this sense is a legal assessment; if in doubt, label it.

According to IHK Lüneburg-Wolfsburg (the regional chamber of commerce, 3 Aug 2026), paragraphs 1 and 4 were not postponed. Only the machine-readable marking under paragraph 2, a provider duty, was postponed: for systems placed on the market before 2 August 2026 it applies only from 2 December 2026.

7. Accuracy in quotes

In the plancraft survey, 26 % named the fear of incorrect results as an obstacle. We found no study of the error rate of AI-generated quotes in the trades. So ask where quantities and prices in the draft come from, your own costing or the language model, and have a person approve every quote. How quotes and invoices can be automated as a whole is covered in Automating e-invoices, quotes and invoices.

What the adoption figures actually say

The range from 4 % to 70 % is not a trend: year, population and method differ. In 2025, Bitkom surveyed a sample weighted on the trades census. The Handwerksblatt article on the plancraft survey states neither the survey period nor how participants were selected, and it includes businesses in Austria. The other three sources do not describe the trades: the September Bitkom release counts companies with 20 or more employees, Destatis reports enterprises by size class, KfW covers the Mittelstand, and construction is not the same as the trades. Destatis notes in a footnote: “According to the EU enterprise definition from reporting year 2025” (our translation). In our assessment, earlier years therefore cannot simply be compared.

We found no current figure from ZDH (the German trades confederation) or the chambers of skilled crafts on AI use for 2025 or 2026, and no source that breaks businesses down by revenue size. So there is no single adoption figure for the trades.

Getting started in three steps

1. Pick one use case. Take the point where things get stuck most today, not the one with the nicest demo.
2. Count. How many calls a day are lost according to your call log, how long does a quote take, how many applications are left waiting? How to turn missed calls into a margin calculation is explained in Missed calls in the trades.
3. A trial with an end date. Set the duration and the success criterion in advance, then decide: keep it or cancel.

How to plan the rollout beyond a single use case is covered in Introducing AI in your business.

Sometimes the better answer is not AI: call forwarding, a telephone answering service or good text templates in your trade software. If nobody in the business has time to maintain the tool, it is better to wait.

Our offer

From here on this is about meine-bots, not neutral criteria. We build AI phone assistants, chatbots and AI agents and connect them to your existing systems. As an add-on, we produce multilingual briefing videos for a newly automated process, in German, Turkish, Polish and Romanian, labelled as AI-generated. Prices and add-ons are on Pricing.

For recruiting and marketing videos without a shooting day, our AI video page describes avatar videos. Our sister site ki-video-marketing.de explains recruiting videos in more detail (in German) and states there itself that such a video does not replace a job interview. An example for trade businesses is there under AI videos for the trades (in German).

The AI Check is the next step after that; it is paid and credited towards a follow-up project. According to its service description it includes a “tool recommendation, even if it does not point to us”.

If you would like to put your figures in context, book a free initial call (30 minutes, a first assessment).

This article is general guidance (as of 20 October 2026) for businesses in Germany and is not legal or tax advice.

Sources

• Bitkom: Digitalisierung des Handwerks, study report, January 2026 (retrieved 20 Oct 2026)
• Bitkom: Erstmals nutzt die Mehrheit der Unternehmen KI, press release of 14 Sep 2026 (retrieved 20 Oct 2026)
• Deutsches Handwerksblatt: plancraft survey, 14 Aug 2026 (retrieved 20 Oct 2026)
• Federal Statistical Office (Destatis): enterprises using artificial intelligence technologies (retrieved 20 Oct 2026)
• KfW Research: Fokus Volkswirtschaft No. 533, 11 Feb 2026 (retrieved 20 Oct 2026)
• Regulation (EU) 2024/1689 (AI Act), EUR-Lex (retrieved 20 Oct 2026)
• Regulation (EU) 2026/1744 (Digital Omnibus on AI), EUR-Lex (retrieved 20 Oct 2026)
• Art. 4 AI Act as amended, buzer.de (retrieved 20 Oct 2026)
• Art. 50 AI Act, artificialintelligenceact.eu (retrieved 20 Oct 2026)
• IHK Lüneburg-Wolfsburg: Digital-Omnibus-Verordnung zur KI tritt in Kraft, 3 Aug 2026 (retrieved 20 Oct 2026)