An assisted living facility can use AI to draft enquiry replies, offer tour slots, send move-in follow-ups and write a family newsletter. A named staff member must still answer anything about a resident's care, health or fees, and personal details stay out of public chatbots. Treat AI as a fast drafter for the front office.
The reason this split works is that most of the work before and after a move is repetitive, but the moments families remember are human: the tour, the first week, the call after a fall. Automate the first kind so staff have more time for the second. The sections below follow a family's journey in order, and each ends with where to hand over.
Who is on the other end of an enquiry
An enquiry rarely comes from the person who will live there. It usually comes from an adult child, often at an awkward time, after a hospital stay or a bad night, and often in a hurry. They write short messages that hide long worries: "Do you have availability? Mum has dementia and is on a lot of tablets." A fast, calm, clear reply matters more than a clever one.
That shapes how you use AI. The assistant is not there to sell. It is there to make sure nobody waits three days for an answer, that the same facts go out every time, and that the right person picks up the delicate part. A good pattern has three layers: an instant acknowledgement, a drafted reply a coordinator reviews, and a clear route to a person for anything touching care needs. For software that offers some of this built in, the tutorial on assisted living software AI features to ask for on a demo lists what to test.
Enquiry replies: draft, review, send
Start by writing your facts once, in a short document a coordinator owns: room types, what the base fee includes, what costs extra, tour days, visiting hours, how an assessment works, what you cannot provide. Load that document into a Project or saved prompt in your chosen assistant, so every draft starts from the same facts rather than from whatever the model remembers. Both ChatGPT and Claude let you share a Project with colleagues: ChatGPT on every plan, Free included, and Claude on its Team and Enterprise plans.
A working prompt, with the enquiry pasted in after removing the sender's surname and phone number:
You are drafting a reply for the admissions coordinator at a small
assisted living facility. Use ONLY the facility fact sheet.
Reply to the enquiry below in under 150 words. Be warm and plain.
Answer what they asked. Offer two tour slots from the list.
Do not comment on the resident's condition or say whether we can meet
their needs: say a short assessment visit will decide that.
If a question is not covered by the fact sheet, say a colleague will
call, and list it under "For coordinator" after the reply.
Enquiry: "Do you have a room free for my father from March? He has
early memory problems and uses a walker. What does it cost?"Illustrative output: "Thank you for getting in touch about your father. We expect to have a one-bedroom apartment available from March, and a studio from April. Fees start from $X a month for the studio, with the care package set after a short assessment visit, so I cannot give an exact figure yet. I can offer a tour on Tuesday at 11:00 or Thursday at 14:30. If you would like to come, I can also arrange a quick chat with our care lead. Which suits you? For coordinator: confirm March availability; family asked about walker access to apartment bathrooms."
What you would fix: the availability claim (the model repeats the fact sheet, which may be a week old), and the line "a quick chat with our care lead", which promises someone's time. The reply avoids saying "we can meet his needs", which is the right call. A person who has seen the assessment should say that, never a draft.
Review is the control. A coordinator reads each draft, checks availability against the live list, and sends. Over a month you will see which answers they change rarely; those are candidates for a template that needs no AI at all.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowTour booking without the phone tag
Tours are the conversion point, and slow scheduling loses families who would have been a good fit. The simplest improvement uses no AI: a booking link with two or three tour days, a calendar that blocks staff time, and a confirmation email. AI adds value around the edges: interpreting a free-text request ("sometime after school run on Thursdays"), proposing slots, and writing the confirmation and reminder wording.
A neat setup for a small facility: a form or booking page captures name, preferred day, who is visiting and any access needs. An automation tool (Zapier and Make are the two common ones) puts the booking in a shared calendar and sends a confirmation. Zapier's free plan covers 100 tasks a month but only two-step Zaps, meaning one trigger and one action, which is enough to drop each booking into the calendar. Adding the confirmation email or a day-before reminder makes it a multi-step Zap, which needs a paid plan; Professional starts at $19.99 a month billed annually for 750 tasks. If you would rather a voice assistant took tour bookings by phone, the tutorial on an AI receptionist that books tours sets out what that costs in a care setting.
Two details prevent embarrassment. First, put the practical details in the confirmation: where to park, who will meet them at the entrance, how long the tour lasts, and that they are welcome to bring the resident or come alone first. Second, send a reminder the day before and one at the right hour. A family visiting for the first time often arrives early, flustered and nervous; a confirmation that names the person they will meet and the door to use removes a layer of stress.
Here is how the free-text step looks in practice. A family writes: "We could do any afternoon next week except Wednesday, but my brother is travelling in from far away so Friday is best." The assistant reads that and proposes Friday 14:30 if it is free, otherwise Thursday 14:30 with a note that a second relative could join by video call. It should offer, not confirm: the coordinator checks the calendar and presses send. The failure to watch for is a slot that clashes with a staff meeting or a resident's activity in the room you use for the tour, which an assistant cannot see unless the calendar is connected.
Think about the hour after the tour as well. A short, personal follow-up within a day is what families remember. Draft it with AI from the coordinator's own notes, kept to facts: "Thank you for visiting. You asked about the garden access and the laundry. Here is the garden plan and the laundry arrangement. The next step, if you want it, is the assessment visit, which takes about an hour." Notes about what the family said about the resident should stay in your own system and out of the prompt unless the family agrees.
Move-in follow-up: the first 30 days
Move-in follow-up is where facilities most often go quiet because the team is busy settling a new resident. A simple schedule catches that: a message on day 1, day 3, day 7, day 14 and day 30, each written by the coordinator's template and personalised by a person.
| Day | Purpose | Who writes it | AI role |
|---|---|---|---|
| 1 | Welcome; who to call; what happens tonight | Coordinator | Polishes wording from a template |
| 3 | Check on settling in; invite questions | Care lead, by phone | None: the call is the point |
| 7 | Photos of activities, if consented; the week ahead | Activities lead | Drafts from a short list |
| 14 | Paperwork still outstanding; care plan review date | Coordinator | Lists what is missing from a checklist |
| 30 | Invite to a short review with the family | Manager | Drafts the invitation |
The point of the table is to show that only some touchpoints should be machine-written. The day-3 call has no AI because a resident's first days are the most delicate, and the family wants a voice. The rest are scheduling and wording, where drafting helps and where a missed one costs trust.
The main worked example, with figures. Say a 40-bed facility has about 3 move-ins a month and the coordinator writes each follow-up message from scratch at roughly 10 minutes each: 5 messages across 3 families is 15 messages, so 150 minutes, or 2.5 hours. With saved templates and AI polishing each one in about 3 minutes, the same set takes 45 minutes. That saves roughly 1 hour 45 minutes a month, and more importantly, the day-14 and day-30 messages stop being forgotten. The saving is small; the reliability is the gain.
The family newsletter without naming residents
A monthly newsletter is the easiest win because it needs no resident data. The content is activities, menus, staff news, upcoming events, a short note from the manager and a reminder of key dates. Ask the activities lead to jot ten bullet points in a shared document. AI turns those into a clean newsletter with a headline, short sections and a call to action at the end.
Before and after. The bullets: "bingo Weds, singer Fri 3pm, new chef from 1st, garden club planting beans, flu jabs 14th, hairdresser moved to Thurs." The draft: "Hello families. March has plenty on. Wednesday is bingo at 2:00, and on Friday a singer will visit the lounge at 3:00. Our new chef starts on the 1st, with a refreshed menu you can see on the notice board. The garden club is planting beans, and visitors are welcome to help. Flu jabs are on the 14th; please talk to the care team if you have questions. The hairdresser now visits on Thursdays." What you would fix: check the "refreshed menu" claim against what the chef plans, and remove "visitors are welcome" if the garden is not open to visitors.
Photos need the most care. A photograph of a resident at the singer's visit is personal information, and a picture of someone with dementia may be shared without their ability to agree. Use photos only where the resident or their representative has agreed in writing to that use, and keep a list. For everyone else, use pictures of hands planting beans, the room set for bingo, or the singer from the back. AI can help write captions, but it cannot decide whether a picture is permitted.
A related piece, on marketing campaigns drafted with AI, covers the public-facing side; the newsletter is private to families, so the rules on consent are tighter.
The fact sheet that keeps every draft honest
Everything above depends on one document, so it deserves more care than the prompts. A fact sheet is a single page, with a date at the top and an owner. Here is an illustrative filled-in version for a 40-apartment facility, with figures invented for the example:
FACT SHEET - updated 3 March, owner: admissions coordinator
Apartments: studio from $X, one-bedroom from $Y a month (base fee)
Base fee includes: meals (3 a day), housekeeping weekly, activities,
24-hour staff on site, emergency call system
Extra: personal care package set after assessment; laundry; hairdresser
Assessment: about 1 hour, free, with care lead; needed before any offer
Tours: Tue 11:00 and Thu 14:30, 45 minutes, family may bring resident
Cannot provide: nursing care requiring a registered nurse overnight
Pets: small pets by agreement, assessed case by case
Do not state: availability dates, fee for care packages, anything
about a named residentThe "do not state" line is the most useful part. It tells both the assistant and the reviewer what must come from a human. Put a review date on the fact sheet and make updating it part of the weekly handover; a draft is only as current as this page. If a family later disputes what they were told, the dated sheet and the saved draft show exactly what was said and when.
A second check worth building in: once a month, ask the assistant five questions the fact sheet does not answer, such as "Do you take couples?" or "Can my mother keep her cat?". The right response each time is "a colleague will call". If it guesses instead, tighten the instruction to answer from the sheet only.
Privacy and consent: what must stay out of a chatbot
Treat anything that identifies a resident, or reveals their health, medication, finances, mood or behaviour, as off limits for a public chatbot. That includes the sender's surname and phone number in an enquiry, which is why the prompt above says to remove them. A consumer AI account may use chats to improve models unless you switch that off; business plans from the main vendors state they do not train on business content by default. Neither makes a resident's file safe to paste. Where you need AI to work on resident data, choose a care-software feature that the vendor documents, with a data-processing agreement your manager has signed.
Consent belongs in two places. First, at admission: the paperwork records who the facility may share information with (a daughter, a son, a representative with legal authority to make decisions), and what kinds. Your newsletter and update templates should draw only from that list. Second, in the privacy notice: say plainly that you use AI to draft routine messages, that a person reads every one, and that no resident information goes into public AI tools. Data-protection and health-privacy law applies to resident information, so have your own adviser check the notice before it goes out.
A practical rule for staff: if you would not read it out loud in the reception area, do not paste it into a chat. Write that on a card by the front-desk computer.
Staff handoff: when the draft stops and a person starts
Most failures in a setup like this are failures of handoff, not of the AI. A family writes a message that sits in a queue because the coordinator is on leave, or a chatbot reassures someone when it should have escalated. Fix this with a short written list of triggers that stop the automation and route to a named person.
- Health or care words: falls, medication, pain, confusion, hospital, "not eating". The message goes to the care lead the same day.
- Money: disputes, late fees, a change in funding. The manager replies; no draft is sent until then.
- Complaint or distress: anger, grief, "I'm worried". A person phones; written replies come later, if at all.
- Safeguarding: any suggestion a resident is at risk. Follow your safeguarding procedure at once, no drafting.
- Silence: any enquiry unanswered after four working hours is flagged to a second person.
An illustrative mistake, and how it would show up: picture a facility whose web chat assistant, set up to answer questions from the fact sheet, was asked "Can you manage my mother's insulin?" and replied "Yes, our care team supports medication needs". The fact sheet did say medication support was available, but not insulin injections, which that facility's staff were not trained to give. The sentence was accurate in general and wrong for that family. It came to light when the coordinator reviewed the weekly transcript. The fix was a hard rule: any question naming a specific condition or medicine gets "I'll ask our care lead to call you today", and the transcript goes to the care lead.
A four-week plan for a small facility
Week one, write the fact sheet and the escalation list, with the manager signing off both. Week two, set up the enquiry draft in a Project and test it on twenty past enquiries, with surnames removed, comparing against what the coordinator actually sent. Week three, put the booking link and tour confirmations live. Week four, write the follow-up schedule and send the first newsletter. Keep a log of every change a person made to a draft; patterns in the log tell you which template wording to fix.
At the end of the month, count three things: time from enquiry to first reply, tours booked as a share of enquiries, and the number of drafts a person had to rewrite fully. If the third number is high, the fact sheet is probably thin or out of date, and the repair is to the document, not the prompt. For a wider view of where a small home should begin, the tutorial on where a small care home should start with AI sets the order, and if you would rather map yours with someone, that is what my AI implementation consultation covers.
One last check that costs nothing: ask a colleague outside the admissions team to read three drafts as if they were a worried relative. If any sentence would make them feel handled rather than helped, the wording needs a person's touch, whatever the draft says.
Questions from assisted living managers
Can AI answer a family's questions about a resident's health?
No. Health, medication, falls and mood are matters for a named member of staff who knows the resident and has the right to share the information. AI can help phrase a message that person has already decided to send, nothing more.
Is it safe to put resident names in a family newsletter prompt?
Avoid it. Write the newsletter about activities, menus, events and staff news, and leave individuals out unless a resident or their representative has agreed in writing. Photos and names need the same permission, whoever writes the draft.
Should the enquiry chatbot give prices?
Only if your price list is simple and current. Many facilities have care levels and assessments that change the figure. A safer pattern gives a starting-from range, explains that an assessment sets the final fee, and offers a call or tour.
Do families need to be told that AI helped write a message?
Not for a spell-checked draft a person has read and owns. Tell them if an AI assistant talks to them directly, such as a chat on your website, and make it easy to reach a human at any point.
Further reads
- AI for Senior Placement Agencies: Family Intake, Matching and Follow-Ups — Family intake and matching, seen from the placement agency's side.
- AI for Apartment Leasing: A Day Plan for Tours, Follow-Ups, Renewals — A tour-and-follow-up day plan from the apartment leasing side.
- How a Small Clinic Can Use AI at the Front Desk Safely — Front-desk AI rules where health information is in the room.
- How Nurseries Use AI to Write Daily Reports for Parents — Daily updates to relatives, written from short staff notes.
- AI for Coworking Spaces: Enquiries, Tours and Room Bookings — Another tour-led business, with enquiry and booking steps compared.
- Supported Living Software: Rotas, Support Plans and Where AI Fits — A plain guide to supported living software: rotas, support plans, daily notes, incident logs, and the jobs where AI adds value and where it should not.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Zapier pricing page; OpenAI Projects help article, Data Controls FAQ and business data page; Claude pricing page (Projects, project sharing, training defaults), all checked 4 November 2026.