It is 9 pm. The mother has been discharged from hospital after a fall, and the daughter sits at the kitchen table with her laptop, looking for a care service. She has questions: which care grade might apply, what does the care cost, how quickly is admission possible? At this hour, no one at the care office can be reached. It is exactly this moment that decides whether a website visit becomes an inquiry or whether the search moves on to the next provider. An AI chatbot on your own care website answers the recurring questions around the clock, collects the request in a structured way and guides care workers searching at the weekend straight to the careers page. This article shows how such an assistant works on a care website, which questions it reliably takes on, how it pre-qualifies requests and where it deliberately hands over to a human.
Key takeaways
- Relatives research in the evening and at weekends: at the end of 2023, 5.7 million (Federal Statistical Office) people counted as in need of care, 86 percent (Federal Statistical Office) of them cared for at home. Inquiries are lost exactly outside office hours.
- The assistant takes on recurring questions about care grade, costs, admission, services, catchment area and open positions using answers approved in advance. It gives no individual care or legal advice and points to a personal conversation instead.
- Instead of an anonymous chat log, a structured request emerges: the concern, the situation and care grade, the region, a callback route with a preferred time window, and consent. Sensitive health data deliberately stays out of the open chat window.
- The same assistant serves care workers: in October 2024 around 17,000 (Federal Employment Agency) positions were registered in elderly care, with a time to fill of 296 days (Federal Employment Agency). It leads to the specific job posting rather than a generic mailbox.
- Handover to a human is defined in advance, for complaints, distressing situations and contract questions. 62 percent (Bitkom) prefer a quickly reachable human contact, and only 50 percent (Bitkom) are satisfied with chatbots.
- Clarify before launch: encrypted transmission, processing in Germany or the EU, defined storage periods and access rights, clear labelling as an automated assistant, plus keyboard operation and contrast per the Accessibility Strengthening Act.
Why No One Is Reachable in the Evening and at the Weekend
The search for a care service usually starts on the internet, and it rarely starts during office hours. At the end of 2023, around 5.7 million (Federal Statistical Office) people in Germany were classed as in need of care, and 86 percent (Federal Statistical Office) of them are looked after at home. In these households, relatives mostly organise care on the side: during a lunch break, late in the evening or at the weekend. Anyone who then lands on a care website with a question meets a phone no one answers, or a contact form whose reply arrives only on the next working day. The gap between the moment of research and the moment of reachability is precisely the gap in which inquiries are lost.
The Numbers at a Glance
Artificial intelligence has arrived in companies too. 36 percent (Bitkom) of companies in Germany use AI, almost twice as many as the 20 percent (Bitkom) a year earlier. Of the companies that use AI, 88 percent (Bitkom) do so in customer contact. In the digital economy the technology is already standard: 98 percent (BVDW) of the surveyed agencies work with generative AI, and 90 percent (BVDW) actively invest in corresponding tools. For a care service this does not mean following every trend, but using the technology where it solves a concrete bottleneck: the lack of reachability outside office hours. How a care service can also become visible in AI-based search services is shown in the article on being found in AI search.
Which Questions an AI Chatbot Answers Reliably
A chatbot for care does not have to do everything. Its value lies in taking on the recurring questions that a care office answers several times a day. These questions can be pre-loaded with checked, approved answers so that the assistant responds in the tone of your care service. The following topics cover a large share of the evening inquiries (project experience).
Care Grade
How is a care grade determined, who carries out the assessment, what do the grades mean? The assistant gives general, checked information and provides orientation without pre-empting an individual classification.
Costs and Financing
What the care fund covers, how a personal contribution arises and which documents are needed are among the most common questions. Clear basics take uncertainty out of the first contact.
Admission and Process
How admission starts, how quickly a first visit is possible and which steps follow can be explained step by step. That way relatives know what to expect.
Services and Area
Which services the care provider offers and whether the address lies within the service area, the assistant clarifies immediately. If the region does not fit, that saves both sides an unnecessary detour.
Open Positions and Applying
Care workers ask about shift models, onboarding and how to apply. The assistant answers this and links straight to the matching job posting and the careers page.
Contact and Appointment
How the care service can be reached and how to arrange a consultation belongs at the end of every dialogue. The assistant leads to the next step without a detour.
The boundary matters: the chatbot gives general, checked information, not individual care or legal advice. As soon as things become specific, it points to the personal conversation or to online booking for the care consultation. How to build a website so that visitors become inquiries in general is explored in the article on more inquiries via the care website.
Pre-Qualify Requests Instead of Only Answering
A good assistant does not only answer questions, it prepares the request for your team. Over the course of a short dialogue it clarifies what it is about, for whom care is being sought and how urgent the situation is. At the end there is no anonymous chat conversation, but a structured request with the key details, bundled in the inbox in the morning. That saves your office the classic phone tag, where both sides keep missing each other and lose time.
- Concern: care for a relative, one's own care, an application or a general question
- Situation: an existing or presumed care grade, time horizon and desired scope of services
- Region: whether the address lies within the care provider's service area
- Contact route: callback number or e-mail and a preferred time window
- Consent: deliberate agreement to being contacted
Data minimisation is decisive. The assistant only asks for what is needed for the first contact, and makes transparent why. Sensitive health data does not belong in an open chat window but in the personal conversation. How to design forms and dialogues in care in a legally sound way is covered on the page on data protection for sensitive health data.
Guide Applicants Straight to the Careers Page
The second audience of a care chatbot is not relatives but care workers. Staff shortage is perhaps the biggest challenge in the sector: in October 2024, around 17,000 (Federal Employment Agency) open positions were registered in elderly care alone, and vacancies stay unfilled for a long time. The average time to fill was 296 days (Federal Employment Agency) in elderly care and 269 days (Federal Employment Agency) in health and nursing care. Anyone looking for staff can hardly afford long response times.
Care workers are on shift. They look for a new employer when they have time, which is often at the weekend or after a late shift, precisely when no one in administration can be reached. An assistant that answers applicant questions about shift models, onboarding and how to apply and links straight to the matching position catches this interest before it cools. Instead of a general e-mail address, it leads to the concrete job posting and the quick application.
One Assistant, Two Audiences
How a careers page convinces applicants is shown in the article on staff recruiting via the care website; how to lower the hurdle of an application is described in the article on the quick application without a cover letter. A chatbot does not replace these pages but acts as the guide that brings interested people to the right place.
When the Chatbot Hands Over to a Human
As useful as an assistant is, it has clear limits, and naming those limits openly builds trust. In sensitive moments people want a person: 62 percent (Bitkom) prefer a quickly reachable, human contact in customer service. Satisfaction speaks the same language: 86 percent (Bitkom) are satisfied with personal contact, but only 50 percent (Bitkom) with the chatbot. For comparison, a well-maintained FAQ page still convinces 60 percent (Bitkom). A chatbot therefore does not replace the conversation, it paves the way to it.
A responsibly set-up assistant therefore recognises when it should not help further, and hands over to your team. These handoff points should be set deliberately, so that no one gets stuck in a loop of automated replies.
- Individual care or health questions that require professional assessment
- Emotional or distressing situations in which a person should listen
- Complaints and urgent matters that need immediate attention
- Contractual or legal questions about services and billing
- Any moment in which the user explicitly asks for a human
Outside office hours the assistant leaves a structured callback request that your team sees first thing the next morning. During office hours it can hand over directly to the phone or to online booking for care. That way the human counselling that is decisive in care is fully preserved, and the hurdle in front of it disappears.
In online retail customer service, the human beats the chatbot: 62 percent prefer a quickly reachable, human contact person.
Chatbot or Contact Form? An Honest Comparison
| Aspect | Static Contact Form | AI Chatbot with Handoff |
|---|---|---|
| Reachability | Reply only on the next working day | Immediate response around the clock |
| Recurring questions | Stay open until a reply comes | Cleared at once with checked answers |
| Pre-qualification | Free-text field, often incomplete | Structured details asked in the dialogue |
| Care workers | One e-mail address for everyone | Direct guidance to the matching position |
| Human counselling | Remains the only route | Deliberately prepared and handed over |
| Effort in the office | Many similar follow-up questions | Routine questions drop away, focus on advice |
Data Protection and Accessibility for Sensitive Matters
As soon as someone writes about the need for care, a care grade or illnesses in a chat, specially protected data arises under the General Data Protection Regulation. For a care chatbot this means: processing belongs in a secure legal framework, ideally on servers in Germany or the EU, transmission must be encrypted, and the assistant should not ask for health data it does not need. Clear consent and an understandable privacy notice are part of this.
Points to Clarify
The second requirement is accessibility. The audience includes older people and relatives under stress; a chat window that only works with a mouse, small type and perfect eyesight excludes some of them. Large buttons, clear contrast, full keyboard operation and labelling that a screen reader can read out are therefore a must, and since the Accessibility Strengthening Act a legal requirement for many providers. How to implement this in practice is described on the page on the accessible care website.
- Encrypted transmission and processing in Germany or the EU
- Data minimisation: no unnecessary health data in the chat
- Clear labelling as an automated assistant
- A visible route to a human contact
- Operability by keyboard, good contrast, legible type
- Defined storage period and access rights for conversations
In Five Steps to a Chatbot on the Care Website
The path to an assistant does not have to be complicated. In practice a structured approach is enough, embedding the chatbot sensibly into the existing care website rather than placing an isolated tool beside it.
- Collect questions: gather the most common inquiries from phone and inbox
- Define answers: store checked, approved answers in the tone of your care service
- Define handoff: set clear points at which the assistant hands over to a human
- Secure data protection: encrypted transmission, EU processing, data minimisation and consent
- Embed visibly: make the assistant available on the home page, service pages and careers page
Whether an assistant can be integrated into your existing site or whether a larger relaunch of the care website is the better occasion depends on the state of your current website. For facilities with fluctuating occupancy, the same assistant can also help to fill vacant day-care places by communicating availability and collecting requests. Which option suits your care service is best clarified in a personal conversation.
Sources and Studies