AI for Senior Placement Agencies: Family Intake, Matching and Follow-Ups

A senior placement agency can use AI to run a structured family intake, turn the notes into a needs summary, build a shortlist from facility data you have checked and dated, and draft follow-up messages. The owner makes every recommendation, discloses how the agency is paid, and keeps family details out of tools that have not been vetted.

A placement owner's day is a stream of emotional calls, each one starting from scratch: who is the person who needs care, what can the family afford, how quickly does a place need to be found, what has already been tried. The calls are the valuable part. The note-taking, the matching spreadsheet and the follow-up emails around them are where AI can help, provided it stays on the right side of one line: it organises what you know, and it does not decide what is best for someone's parent.

Follow me on Instagram@sagnikteaches

Where the hours go in a placement

Before adding software, spend a week noting where time goes on three or four live cases. The pattern is usually the same, and it sets your priorities.

Connect on LinkedInSagnik Bhattacharya

Step one: an intake script that never misses the basics

A script is not a way to make calls robotic. It is a safety net so you remember the questions that matter when you are listening to a worried daughter or son. Write it once, keep it on screen, and let your own judgement lead the conversation. Keep a distinct section for each of the topics below.

Subscribe on YouTube@sagnikteaches

The intake structure

• Who is calling and who is the person needing care. Relationship, who has authority to decide, who else is involved. Ask early whether the person has been part of the conversation.
• Current situation. Where they live now, what is not working, what triggered the call, whether there is an urgent deadline such as a hospital discharge.
• Care needs, in the family's words. Help with daily activities, mobility, memory, medication, any condition the family chooses to share. Record what they say; do not interpret clinically.
• Preferences. Location, size and feel of a place, pets, faith or community, food, a room of their own.
• Budget and funding. Monthly budget range, sources of funding, anything pending. Ask gently and explain why.
• Timing. Needed date, flexibility, whether a short stay or trial is possible.
• Anything they have already tried or visited.
• How you are paid. Say it in the first call, before gathering the detail (see below).

Say how you are paid, early and plainly

Many placement agencies are paid by the facilities they refer to, not by the family. That is a legitimate model, and it is also a conflict of interest if a family does not know. Disclosure is good practice and, in some places, a legal duty, with rules on telling clients how the fee is worked out, when it is paid and, if they ask, how much it is. Ask your adviser what applies to you, then build it into the script. An illustrative form of words:

Before we start, I'd like you to know how this works. Our service is free to families. Where a family moves into a community we have introduced, that community may pay us a fee. That is why I'll always show you options that suit your needs even if they don't pay us, and I'll tell you which ones do. You are free to choose any place, including ones we are not paid by.

Put the same statement in writing in your welcome email and on any shortlist you send, with a marker against each facility showing whether you have an agreement with it. Never let fee level influence which places go on a shortlist or the order they appear in. An AI tool must never see compensation data when it prepares a shortlist, so that it cannot be shaped by it.

Step two: from notes to a needs summary

After the call, you have rough notes. AI can turn them into a clean needs summary in a standard layout. It must not add needs, diagnose, grade the person's condition or decide a level of care. It restates what the family said.

Turn these call notes into a one-page needs summary for a senior
placement adviser. Use ONLY what is in the notes.

Sections: Situation / Care needs as described by the family /
Preferences / Budget and funding / Timing / Open questions

Rules:
- Quote the family's own words for care needs.
- Do not add medical interpretation or suggest a level of care.
- If something important is missing, list it under Open questions.
- Do not include names, addresses or exact dates of birth.

NOTES
[paste notes with identifying details removed]
END

Here is an illustrative result from messy notes:

Situation: Parent lives alone; the family is worried after two recent falls. Hospital discharge expected within three weeks. Care needs as described: "needs help getting dressed and with medication; forgets meals." Preferences: small place, near the daughter, pets allowed if possible. Budget: family suggests up to a monthly figure they will confirm. Timing: discharge date drives the move. Open questions: level of care assessed by the hospital team; who holds power to make decisions; funding sources not yet discussed.

The Open questions line is the most helpful part. It turns what you forgot to ask into a short list for the second call. Check the draft against your own notes before it goes into your records, and delete anything the model added that you did not say.

Step three: the shortlist, from facility data you have checked

This is the step where AI is most tempting and most dangerous. A chatbot asked "which care homes in this area have availability and cost under X" will happily produce an answer, drawing on out-of-date or invented details. Never use it that way. Your shortlist comes from your own facility database, built from information you have verified, and AI only helps with sorting and writing.

Build the facility record

For every facility you work with, keep a record with the same fields and a date and source next to every fact. A spreadsheet is enough.

Two rules keep this honest. Every availability and price carries a date, and anything older than a stated number of days, such as 14 for availability, is marked stale and must be rechecked by phone before it goes to a family. And the fee agreement lives in a separate sheet that the matching step never reads.

Matching: criteria first, then evidence

Take the needs summary and write the family's criteria as a short list: must-haves, such as a unit for memory support, a room available within the family's timeframe and a fee within budget, and preferences, such as nearness to a daughter or pets. Then filter your database against the must-haves yourself, or with a spreadsheet filter. AI can then help lay out the remaining options as a side-by-side comparison.

Using ONLY the facility records below, build a comparison table for a
family. One row per facility, one column per family criterion.

Rules:
- Copy facts exactly, including the date and source of each.
- If a fact is missing or older than 14 days, write RECHECK.
- Do not recommend, rank or score facilities.
- Do not mention fee agreements or commissions.

FAMILY CRITERIA
[paste]

FACILITY RECORDS
[paste records only; no fee-agreement data]
END

The output is a table for you, not for the family, until you have rechecked every RECHECK. You then pick two to four options for the shortlist, and you write, or edit, the commentary on each: what you know from visits and conversations that is not in the data. That commentary is your professional value, and it is what a family is paying for, even if they never pay you directly.

The chatbot shortlist nobody could trace

In this illustrative case, an adviser asked a general chatbot to suggest homes near a family's preferred area that had a memory unit. The reply listed four, with confident detail about fees and a "recently rated highly" note. One no longer offered memory care, one had closed a wing, and the "rating" could not be traced to any report. The adviser caught it only by ringing the first home. The rule that came out of it was that nothing reaches a shortlist unless it is in the dated facility database, however plausible it sounds.

Step four: follow-ups that help a family decide

Families rarely decide in a single call. They tour, discuss, worry and sometimes disappear. Well-judged follow-ups help, and drafting them is something AI does well, because each is a short, personal message with a clear next step. The danger is tone: a family under stress will notice pressure instantly, so the prompts should ask for warmth and options, never urgency.

Here is an illustrative message after a tour, drafted from a prompt that supplies the family's stated priorities and asks for a short, kind note with a list of questions:

Hello [first name], thank you for visiting [facility] yesterday. I know visits like this can bring up a lot. If it would help, here are three things families often like to ask a second time: how staffing works at night, what is included in the monthly fee and what is charged as extra, and how the home keeps families informed. Would you like me to arrange a second visit, or look at one of the other options again? There's no rush, and I'm here whichever way you are leaning.

Before sending, check that it makes no claim about the home you have not verified, that it contains no pressure, and that the tone fits what the family has told you. If they seemed tired and overwhelmed, a phone call may be kinder than any message. Draft with AI, send as yourself, and never auto-send follow-ups to a family in distress.

A tour question list built from what the family said

Tours go better when the family arrives with questions tied to their own worries, not a generic list. Give the model the needs summary and ask for eight questions to put to the facility, grouped by topic, using only the family's stated concerns. For a family worried about falls, forgetting meals and a parent who dislikes crowds, an illustrative result might include: how are residents checked on overnight and how often; how are meals and fluids monitored for someone who forgets to eat; what happens after a fall, and who is told and when; how many residents share a dining area at once; and what is charged on top of the monthly fee. You then add two questions from your own experience of that particular home. The family leaves with a page to write answers on, and you gain a record to compare tours fairly.

When family members disagree

Siblings often want different things, one prioritising price and another nearness or quality. An AI summary can help by listing each person's stated priorities side by side, drawn from your notes, with no judgement about who is right. Share it with the family as a discussion sheet. The point is to make the disagreement visible and specific, so the conversation moves from "you never listen" to "we rank price and distance differently". Do not let the model suggest a compromise or a winner; that remains a human conversation, with the older person's own wishes at its centre wherever they can express them.

Privacy: a stricter standard for a vulnerable group

Intake records contain health, family and financial information about older people, some of whom cannot give their own consent. Handle them accordingly.

• Collect the minimum. Record what the placement needs, not a full medical history.
• Keep identifiers out of general AI tools. Use initials or "the parent", and no addresses, dates of birth or account details.
• Check vendor terms before connecting any tool to your CRM or phone system: where data lives, whether it is used for training, who can see transcripts. The privacy-terms checklist lists the questions. Take advice on the rules that apply to you.
• Get consent for recordings. If a call is recorded or transcribed, say so at the start.
• Share with facilities only what is needed, and only with the family's agreement.
• Delete on a schedule, including notes in AI tools, once a case has closed.

Placing a father within three weeks of hospital discharge

Picture a two-adviser agency taking a call from a daughter whose father will leave hospital in three weeks and cannot return home. Times are illustrative.

1. First call (45 minutes). The adviser follows the script, states how the agency is paid, and takes notes by hand. Afterwards, ten minutes of AI tidying yields a needs summary with four open questions, including who holds decision-making authority and the hospital's care assessment.
2. Second call (20 minutes). The adviser covers the open questions and confirms a monthly budget range. The summary is updated.
3. Filter and compare (30 minutes). The adviser filters the facility database for homes within the area that offer the required care level and expect a room within four weeks. Seven remain, three of which have availability data older than 14 days. The comparison table marks them RECHECK.
4. Verification calls (60 minutes). The adviser rings the three, confirms availability and fee details, and updates the record. One has no room after all, so it drops out.
5. Shortlist (30 minutes). Three homes go to the family, each with the adviser's own two-sentence comment from past visits, a clear mark showing which have fee agreements with the agency, and the dated facts.
6. Tours and follow-up. AI drafts a tour question list and a day-after message for each visit; the adviser edits and sends them.

The adviser's time on the desk-based parts, about two and a half hours here, would have been closer to four without the structured summary and comparison. Treat those numbers as a guide and measure your own. None of the steps involved AI deciding anything about the father's care.

Eight rules for every adviser on the team

1. AI organises and drafts. People decide, recommend and send.
2. Shortlists use only the dated, verified facility database.
3. Every availability and price is rechecked by phone before a family sees it.
4. Fee agreements are disclosed early and in writing, and never fed to AI or used to order a shortlist.
5. No names, addresses or dates of birth in general AI tools.
6. No clinical interpretation or level-of-care judgement from AI.
7. Follow-ups never create urgency.
8. Family and facility data are deleted on schedule.

Five measures of an accurate, fair placement desk

• Accuracy: how often a RECHECK turns out to be wrong, and how many shortlist facts a family or facility corrected.
• Speed: minutes from first call to a sent shortlist, before and after.
• Completeness: how many second calls were needed to fill missing intake answers.
• Tone: read a sample of follow-ups monthly for pressure, and ask families for feedback after a placement.
• Disclosure: confirm every shortlist carries the fee statement and the markers.

The similar structure in enquiry to fact find for mortgage brokers shows how another advice business handles the same pattern, and where a small care home should start with AI shows the other side of the table. If you would like to map your own intake and shortlist process with someone, that is what my AI implementation consultation covers.

Questions senior placement owners ask about AI

Can AI pick the right care home for a family?

It can help you sort options against needs the family has stated, but it cannot know what suits a particular person. The choice depends on care needs, budget, location, feel of the place and many things only a visit and a conversation reveal. Use AI to prepare a shortlist from checked data; you and the family decide.

Where should facility data come from?

From the facility itself, confirmed by you and dated, and from official inspection or licensing reports where they exist. Never let an AI tool fill in prices, vacancies, care levels or ratings from its own memory or from unverified websites. Store the date and source for every field, and re-check before sending any shortlist.

Do we have to tell families how we are paid?

Being open about it is good practice and, in some places, a legal requirement. Many placement agencies are paid by the facility, not the family, which can create a conflict of interest if not explained. Ask your own adviser what applies to you, and put a plain statement in your first conversation and your written materials.

Is it safe to put family details into an AI tool?

Only with care. Intake records contain health, financial and family information about vulnerable people. Keep names and identifying details out of general AI tools, use tools whose data terms you have checked, and take advice on the privacy rules that apply to you. Collect the minimum, and delete what you no longer need.

Can AI write follow-up messages to families?

Yes, and it is one of the safest uses, because a person reviews and sends them. Keep them kind, specific and free of pressure. Never use AI to create urgency or to steer a family towards a facility that pays you more; the follow-up exists to help them decide, not to close a sale.

Further reads

• How to Automate Buyer and Vendor Follow-Up With AI — Follow-up automation for a long, emotional decision.
• How to Clean Up Customer Records Before You Add AI — Tidy your records before adding AI to them.
• AI Client Intake for Law Firms: Qualify Enquiries Out of Hours — Qualify and capture enquiries out of hours.
• How Wedding Venues Use AI for Viewings, Enquiries and Follow-Ups — Viewings and follow-ups for a venue business.
• AI Answering Service for Home Care Agencies: After-Hours Call-Outs — After-hours call rules for a care business.
• AI for Nanny Agencies: Family Briefs, Candidate Shortlists and Check-Ins — Family briefs and fair shortlists for a placement agency.
• AI Governance for a Small Business: Who Decides, Approves, Checks — Who decides, approves and checks AI output in a small firm.
• AI for Assisted Living Facilities: Enquiries, Tours, Family Updates — Draft enquiry replies, book tours, follow up after move-in and write family newsletters with AI, while keeping resident details and care decisions human.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: a large placement service's statement that participating communities pay it, and a referral-agency statute's fee-disclosure section (both checked 4 November 2026); general placement-agency practice. Fee disclosure, privacy and licensing rules vary and must come from your own adviser and regulator; facility data must be confirmed with each facility and official inspection reports.