In short: Today, AI and automation mainly take office work off the hands of trade businesses. Enquiries are taken and recorded, quotes prepared, invoices created from the job, payments matched and reviews requested. On site, the trade remains a trade, and a person still decides on anything that commits the business.
Automation and AI are not the same thing
Automation means that one step triggers the next, according to a rule you have set. When the job is completed, the invoice is created. When it is three days overdue, a reminder goes out. This works reliably and without any artificial intelligence at all.
AI comes in where rules are not enough: understanding what a caller needs, reading an incoming invoice, drafting quote text from photos and keywords. The best workflows combine the two. The AI understands and prepares, and the automation carries the result forward.
Five places where it pays off today
Taking enquiries
An assistant answers calls while you are up on the roof and also reads emails and forms. It asks for the address, what the customer needs and their preferred date, and creates a complete enquiry. To see how this works step by step, look at the workflow from enquiry to quote.
Preparing quotes
A photo, voice note or site measurement becomes a draft with your line items. If your suppliers' price lists are connected, the draft is calculated with current prices. You check and approve.
From job to invoice
What is recorded on site, meaning times, materials, photos and signature, is the same information the invoice is created from. Nobody retypes anything. The whole route is shown in the workflow from job to invoice.
Payments and accounting
Incoming payments are matched to the invoice, unpaid items flag themselves, and receipts go to your tax adviser continuously. This requires an integration with your bank and accounting.
Reviews and reference projects
Once the job is completed, a request for a review goes out, and the job photos become a draft for a reference project on your website. Nothing is published until you approve it.
Where the limits are
AI does not make professional judgements and does not accept liability. It can be wrong and still sound convincing. Three things therefore belong in human hands:
- Prices and commitments that bind your business
- Technical decisions, for example on the safety of an installation
- Information about standards and regulations
A good system shows openly what has been prepared and waits for your approval at these points.
What to bear in mind legally
This section gives an overview and is no substitute for legal advice.
Data protection. As soon as a system processes names, addresses or phone numbers, you need a data processing agreement with the provider under the GDPR (German: DSGVO). Clarify where the data is processed, how long conversations are stored and who can access them.
EU AI Act. People should be able to tell that they are speaking or writing with an AI. A phone assistant therefore introduces itself as a digital assistant. Since February 2025, companies have also had to make sure that employees who work with AI are adequately trained.
How to get started
- Choose one workflow. The most frequent and most tedious one first.
- Collect examples. Ten typical enquiries, five old quotes, your price list.
- Start with approval. In the first few weeks, nothing goes out without you looking at it.
- Review after a month. What went well, and where did the system get it wrong? Only then does the next workflow follow.
To see what difference this makes day to day, look at the comparison before and after. The building blocks behind it are described on the pages Automation and AI assistants.
Frequently asked questions
What is the difference between AI and automation?
Automation carries out fixed rules: when the job is done, the invoice is created. AI takes on what cannot be put into fixed rules: understanding a phone call, reading a photo, drafting a text. In practice the two work together.
Does my business need its own IT skills for this?
No. You show how you work today and provide examples. The provider takes care of setup, operation and adjustments.
Where should a small business start?
With a single workflow that comes up often and takes a lot of time. Usually that is taking enquiries or getting from the finished job to the invoice.

