A call comes in while reception is on the other line. A WhatsApp message arrives at 10 pm. Someone sends a photo of a window and asks what it costs. Reply late, and the enquiry often goes to the next provider. Our AI assistants reply straight away. You decide in advance what they may say live. Everything else comes to you as a draft.
The three businesses are made up. The workflows are the ones we set up. The replies in the examples are written by hand, and no AI runs behind them.
On the phone and in chat, it answers questions about opening hours and directions. When someone mentions symptoms, it refers them to the practice. Callback requests land in the inbox with name and number.
See the Lumen practice → WhatsApp and email · Developers, estate agentsThe assistant reads the enquiry and writes a reply with the price and the available flats from your list. Your adviser approves it with one tap. Anyone still waiting after a day goes to the top of the list.
See Alba Wohnen → Photos and measurements · TradesThe customer sends photos and measurements. The assistant checks that they match, and your price table works out the range. You approve the estimate, together with a date for measuring up.
See Lamella Sonnenschutz →Phone and chat reply straight away, but only from a knowledge list you sign off in advance: opening hours, address, procedures, free appointments. Promises, diagnoses and prices outside your list are not on it.
The assistant writes every WhatsApp or email reply and every estimate as a draft. They go out when you or your team tap “Approve”.
Article 50 of the EU AI Act requires this at the first interaction at the latest. We go further: on the phone in the first sentence, in chat in the first message. People who know who they are talking to ask clearer questions.
We sign a data processing agreement (Art. 28 GDPR) with you, and it names the AI services we use. The assistant does not ask for health data and never writes to anyone who has not contacted you first.
Most of an assistant is a fixed workflow: take the enquiry, sort it, send reminders, file it. Ordinary code does that. It works the same way every time and costs almost nothing. The AI only comes in where a message has to be understood or a reply written. That keeps running costs low, and every mistake can be traced.
Price and status. We don't quote a fixed price yet. We are setting up the first assistants now and have no case studies to show. After a short call you get a written quote for your business.
It depends on how many calls come in and what it should answer. We give you the price in writing after a short call. There are no fixed package prices yet, because the first assistants are being built now.
In principle, yes. It must say that it is an AI at the first interaction at the latest (EU AI Act, Art. 50), tell callers how their data is processed, and you need a data processing agreement (Art. 28 GDPR) with the provider. Because you are bound by medical confidentiality, the provider must also be bound to secrecy (§ 203 German Criminal Code).
Our assistants don't ask about symptoms. If someone mentions symptoms, they refer them to the practice or to the out-of-hours medical service on 116 117, and in an emergency to 112. The legal review for your practice stays with your lawyer.
No. It only replies to people who have contacted you, and only about that enquiry. It does not send advertising to old contacts. Without consent, that is generally not allowed under German unfair competition law (§ 7 UWG).
It says so and takes a callback request. It doesn't guess and doesn't make up prices. The question comes to you, and we add the answer to the knowledge list if it comes up often.
No. It answers when everyone is busy or you are closed, and it writes drafts. Your team makes the decisions.
Tell me which enquiries go unanswered at your business. I reply within 24 hours on working days.