AI Chatbot for Senior Care: Answer Families, Not Residents
The person messaging your website at ten on a Sunday night is rarely the resident. It is a daughter, a son, or a spouse, usually worried and usually researching several places at once. A chatbot can answer them well, as long as you draw one hard line around anything to do with a named person.
The short version
- Your enquirer is a family member. Write every answer for a stressed relative, not a resident.
- Named residents are off limits. Any message about a specific person goes straight to staff.
- Evenings and weekends are the point. That is when families research care.
- Answer the general, escalate the personal. That single rule covers almost every decision.
The enquiry you are missing arrives at 10pm
Families do not shop for care during office hours. They start after the hospital calls, after a fall, or after a visit that made things obvious. The searching happens late at night and at weekends, when the house is quiet.
Your website is open at that hour. Your admissions team is not. A contact form is a request for a callback, which is a slow answer to an urgent-feeling question. Many people will read three provider websites, get answers from none, and go with whoever replies first on Monday.
A chatbot changes that evening. It answers the questions your site already covers, takes the details for a tour, and says when a human will call. That matters a lot to someone who has just spent an hour worrying. The same after-hours pattern runs across service businesses, and we cover it in our after-hours support guide.
The line the bot must never cross
This is the part to settle before anything else. A senior care chatbot must never discuss a named resident. Not their health, not their care plan, not their medication, not their mood, not whether they ate lunch, not their bill. It should not confirm whether a particular person lives with you at all.
The reason is simple. The chat window has no idea who is typing. A caring daughter and someone with no right to that information look identical to a text box. So the rule is not "be careful", it is "stop". The bot hands over to staff and shows the phone number.
Write that instruction into the setup, then test it. Ask the bot about a made-up resident from five different angles before you launch. Our guide to chatbot guardrails walks through how to phrase and check rules like this.
| A family member asks | What the bot does | What it needs |
|---|---|---|
| What levels of care do you offer? | AI answers in full | Your published descriptions, in plain words |
| What is included in the monthly fee? | AI answers in full | Fee structure page, plus what is billed separately |
| Can I book a tour on Saturday? | AI books or takes details | Tour slots, or a request queued for Monday |
| Do you have a room available now? | AI takes details, staff confirm | Never quote availability the bot cannot verify |
| How is my mother doing today? | Stop and escalate | Named resident, staff only, phone number shown |
| I want to raise a concern about care | Stop and escalate | Straight to a manager, with your complaints line |
Notice the fourth row. Room availability changes hour to hour, and a wrong yes costs you a family's trust on the first contact. Let the bot collect the details and promise a call, rather than guess.
Three different conversations, one widget
Senior care sites carry three audiences at once. New families are an admissions conversation. Current families are a manager conversation, often an urgent one. Job seekers are a recruitment conversation, which matters given how constantly care teams hire.
One early question sorts them: are you looking into care, are you already family here, or are you asking about a job? Each answer routes somewhere different. Replies land in Slack or Microsoft Teams on every plan, Free included, which suits managers who spend the day walking the building rather than sitting at a desk. The mobile app is on every plan, including free.
Tours, waiting lists, and the honest answer
Booking a tour is the one action worth pushing for, because it turns research into a decision. The bot can take a name, a phone number, a rough timescale, and the level of care the family has in mind. Your admissions lead then starts Monday with a real conversation instead of a voicemail tag.
Waiting lists deserve the same honesty as availability. The bot can explain how your list works and how families are contacted, but it should not estimate a wait. If your site publishes a figure, the bot repeats it. If not, staff will explain, and the bot says when they will call.
Realistic volume, and what it costs
A single community sees a modest number of website chats a month, not thousands. Enquiry volume in senior care is low and high value, the opposite of retail. So the free plan covers a lot of providers, and the real question is seats and channels rather than volume.
| Plan | Price | Messages a month | Why a care team picks it |
|---|---|---|---|
| Free | $0 forever | 500 | 1 chatbot, 2 seats, Slack, Teams and WhatsApp, mobile app |
| Starter | $25/mo | 5,000 | 3 agent seats plus the owner, 3 chatbots, workflows and outbound campaigns |
| Professional | $95/mo | 25,000 | 10 agent seats plus the owner, 10 chatbots, AI Insights and AI Email |
| Enterprise | Talk to us | Custom | Seat volume, invoicing and a data processing agreement |
Groups running several homes can give each site its own chatbot: Starter at $25 covers three and Professional at $95 covers ten. Annual billing takes 20% off either, and the full breakdown sits on the pricing page. Larger groups with procurement teams should ask about the Enterprise tier.
A setup week that works
EzyConn trains by crawling your website, so the fee structure, care levels and visiting information you publish become answers in minutes. GPT-4o and Claude are on every plan, and replies come back in many languages, which matters when a family member abroad is part of the decision.
Spend the rest of the week on three things. Write the escalation list, meaning every phrase that should stop the bot cold. Agree who receives handovers in the evening and at weekends. Then read two weeks of transcripts with a manager who knows families well, and fix the answers that sound wrong.
Frequently asked questions
Can an AI chatbot answer questions about a specific resident?
No, and it should be set up so that it cannot. The moment a message names a resident, the chatbot should stop answering and pass the conversation to staff with the phone number on screen. Care plans, health updates, medication, falls and billing for a named account all belong to the people who know that person.
What should a senior care chatbot actually be allowed to answer?
The published, general things. Levels of care and what each one means, what the fee includes, visiting hours, parking, meals, pet rules, how tours are booked, what happens on move-in day, and how to apply for a job. These questions repeat every week, and none of them touch anything private.
Why do so many family enquiries arrive outside office hours?
Because that is when families have the conversation. People research care after a hospital call, after a difficult visit, or late at night when the house is quiet. Weekends are busy for the same reason. If your website only answers between nine and five, those enquiries wait until Monday, and some go elsewhere.
How do we stop the chatbot from sounding cold to an upset family member?
Two settings do most of the work. Keep answers short and warm rather than cheerful, and make the route to a person obvious in the first reply. A family member choosing memory care for a parent does not want a sales pitch. They want a clear answer and a name to call.
What does this cost for a single community?
The free plan is $0 forever and covers 2 seats, 500 messages a month and one chatbot. That is enough for many single sites. Paid plans are Starter at $25 a month and Professional at $95 a month, with 20% off on annual billing.
Be awake when families are
The free plan is $0 forever, with no credit card: 2 seats, 500 messages a month, one chatbot on your site. Enough to see which questions your website is not answering at 10pm.
Start freeLast updated . EzyConn plan details are current as of August 2026. This article is general guidance on setting up website chat, not advice on your regulatory obligations. View more guides.