Homes First
Proposal · nursing-home matching platform · Petaling Jaya, Selangor

Homes First

A phased route to a nursing-home matching platform that starts by selling homes something they already want, and builds the directory as a by-product.

Prepared for Choong
By Eugene Koo · SoftSupplies
9 September 2026 · Draft for discussion

The end goal is right. The first step is not an app.

What we agreed in the meeting, written down so we can hold each other to it.

The goal is clear: when a family needs a nursing home for a parent, they come to one place and get matched to a home that can actually take them. Nobody disagrees with that. The question is the order of operations, and the order decides whether the platform ever has anything in it.

A matching platform lives or dies on the supply side: homes, with accurate and current information about what they can take. Families are the easy half. They arrive stressed, once, and want an answer this week. Homes are the hard half. They have no reason to fill in a profile on a directory that has no families on it yet, and even less reason to keep it up to date.

So this proposal sequences the build around four positions.

POSITION 1

Do not start with an app

Nobody searching for a nursing home downloads an app to do it. The need is occasional, stressful and one-off. An app is the wrong first surface for the demand side, and the most expensive one to build.

POSITION 2

Families get a WhatsApp chatbot

Meet them where they already are. A number they message, a short conversation, an introduction to a suitable home. No install, no account, no learning curve, and it works on the phone they already hold.

POSITION 3

Onboarding homes is the real problem

A two-sided marketplace has a cold-start problem, and here it sits squarely on the supply side. A home will not maintain a listing for a platform that has not yet sent it a single family.

POSITION 4

Sell homes WhatsApp automation as the wedge

Sell homes a tool they want and will pay for. Onboarding to the platform becomes a side effect of the sale, and the data the tool produces every day is exactly the data matching needs.

How the sequencing beats the cold start

Supply pays to be onboarded. The data stays current because the home uses the tool to run its day.

The usual directory has two failure modes. Either homes never sign up, or they sign up once and the listing rots: beds shown as available are full, a home that stopped taking dementia cases is still listed for dementia care, the contact person left last year. Families call, get turned away, and stop trusting the platform.

The wedge fixes both at once. A home buys SS Care because it wants families to stop phoning the ward and wants records with photo proof. To use it, the home registers its residents and its staff, and carers record vitals every shift. From that point on the platform knows, without anyone filling in a form:

  • Residents in care right now, and therefore occupancy against licensed capacity.
  • Care capabilities, inferred from what is actually recorded: glucose monitoring, oxygen saturation tracking, and as further record types are added, tube feeding, dementia routines, mobility support.
  • Staffing: how many registered carers, on how many shifts, per resident.
  • Responsiveness: how quickly the home replies to an enquiry, measured from the same WhatsApp number.

The home never "updates its listing". It just runs its day, and the listing follows.

A home that is not ready to buy the tool can still join by a light path: one WhatsApp message to the Homes First number when a bed opens or fills. It costs the home a little, and its listing says exactly how fresh it is.

Home buys SS Care carers record vitals daily family group gets the card Live operational data residents · occupancy capabilities · staffing Family chatbot matches location · budget · needs introduction or visit MATCHED FAMILIES BECOME NEW RESIDENTS · THE HOME KEEPS USING THE TOOL · THE DATA STAYS CURRENT PHASE 1 · PAID BY THE HOME BY-PRODUCT · NO FORMS PHASE 2 · DEMAND SIDE

The platform's supply side is paid for by the supply side. The demand side is added only once there is something to match against.

The order, and what unlocks each step

Nothing later starts on hope. Each phase has a condition the phase before it has to meet first, so the expensive work is never begun on optimism.

1

Sell automation to homes

Homes buy SS Care and run their day on it. Every sale is one home onboarded, with live operational data as a by-product.

Starts now · the home pays
Unlocks phase 2 N homes live across enough of PJ
2

A WhatsApp chatbot for families

One public number. A family answers a few questions and is introduced to a home that can actually take their parent.

After the first homes are live
Unlocks phase 3 N homes and M enquiries a month, two months running
3

Directory, console, and only then an app

Public pages with availability that is true today, and an enquiry console for homes. The app, if it is built, is for homes — never for families.

Gated · not started by design

The two gates are the whole discipline of this plan. N and M to be agreed with Choong

1

Sell automation to homes

The wedge · starts now

SS Care is the product. Each sale is one home onboarded with live operational data, whether or not the word "platform" is ever mentioned.

What a carer and a family actually see, one shift, about twenty seconds end to end:

Carer's phone · SN Sheila

This is the live product, not a design. Tan Ah Kow is a demonstration resident.

What gets built

Very little that does not already exist. SS Care is built, deployed, and verified end-to-end with a real nursing home. What a carer does today:

  • Sends /care to the home's WhatsApp number. A WhatsApp Flow (Meta's in-chat form) opens. Picks a resident, enters blood pressure, heart rate, SpO₂, temperature, and optionally glucose.
  • Photographs each instrument, camera only, up to six photos. Taps Save. Roles (carer, doctor, admin) are enforced on the server, not the phone.
  • Within about 30 seconds a card appears in the resident's existing family WhatsApp group. At 21:00 the family gets a daily summary. Doctor's notes and photos are deliberately never sent to families.
  • A vision model reads the monitor in each photo and compares it with what was typed, flagged in the admin dashboard as agrees, differs, or unreadable. It never overwrites the carer's number; a person stays accountable for the record.
  • The home's admin runs everything else from a web dashboard: records with photos and CSV export, residents and their family-group binding, staff and roles, and the WhatsApp connection. Shown below.

Work needed for Phase 1 to scale beyond one home: multi-tenant hosting (one instance per home today), a hardened bot supervisor and re-linking path, backups and a photo-retention policy, and an onboarding runbook for Meta business verification.

Who pays

The home. A monthly subscription per home pricing to be decided with Choong, possibly tiered by resident count. Setup (WhatsApp Business registration, staff and resident onboarding) as a one-off to be decided.

The point is that the home pays for a tool, not for a listing. Platform participation is included in the price and never sold separately.

Data it produces

Per home, continuously: resident roster and count; registered staff, roles and shift coverage; the record types in daily use, which is how care capability is inferred; time-of-day activity; and the home's own WhatsApp number, already verified with Meta, ready to receive enquiries in Phase 2.

What "done" looks like

The first home recording real residents on SS Care every day (not a demo). Then a second and third home onboarded from a runbook rather than by hand, each on its own verified WhatsApp Business number, each with a supervisor-managed bot and nightly backups. At that point Phase 1 is a repeatable sale.

Risk that stops it

Meta business verification is the slow gate for every new home. It is an account-state problem, not a code problem, and lead time is outside our control. Onboarding should start the verification on day one and run staff training in parallel, so it never becomes the critical path twice.

Why a home buys

Families stop calling to ask “how is Mum”

The answer is already in their group, every shift, with the evening summary at nine.

Records with photo proof

Every reading carries the instrument photo and a second opinion from the vision model. Disputes about what was recorded go away.

Nothing to install, nothing to train

Carers already use WhatsApp. The form opens inside the chat they have open all day.

The home's side: the admin dashboard

Carers never leave WhatsApp; the home's admin gets a web dashboard. It is built and in use for the current deployment. Records with the instrument photos and the vision check beside each reading, CSV export, residents bound to their family groups, staff and roles, and the WhatsApp connection page.

AdminSS Care admin dashboard (web)Built · shown with sample data
SS CareGolden Years Careadmin · Sister Aida
RecordsResidentsStaffWhatsApp
Wed 9 SepAll residentsAll shiftsPhoto check: 3 differs this weekExport CSV
TimeResidentBPHRSpO₂TempCarer · shiftPhotosPhoto checkFamily
1:25 pmTan Ah Kow · R3A145/807297%36.6SN Sheila · PMagreessent 1:25 pm
1:22 pmWong Mee Lan · R5B128/766898%36.4SN Sheila · PMagreessent 1:22 pm
1:18 pmAbdul Rahman · R7A152/888195%37.1SN Sheila · PMdiffers · BP read 132/88sent 1:18 pm
8:10 amTan Ah Kow · R3A138/787098%36.4SN Aida · AMagreessent 8:10 am
8:06 amLim Siew Hong · R2B119/706497%36.2SN Aida · AMunreadable · glaresent 8:06 am

"Differs" does not change the record; it flags it for the admin to look at the photo. The carer's number stays the number of record, and a person stays accountable. Residents and staff names are sample data.

See it working now. Scroll-driven walkthrough at homesfirst.softsupplies.com.my/demo.
Ask for a live demo on WhatsApp — Eugene registers you as staff for the session and you record a reading yourself.

The light path: homes that do not buy SS Care

Not every home will subscribe on day one. A home can still be listed on Homes First with nothing to install and nobody to train: it messages the Homes First WhatsApp number when a bed opens or fills, and the listing shows what the home last confirmed, and when.

What the home pays. A small fee per day listed, or a light monthly subscription price to be decided with Choong. Paying, even a little, is what separates a home that means it from one that filled in a form once.

What the platform gets. Declared availability with a timestamp. Weaker than an SS Care home's live count, and shown as such: families see "confirmed by the home Tue 9:40am", not a bed count that pretends to be live.

  • A listing that stops replying drops out of matching, so nobody is introduced to a bed that is not there.
  • Every light-path home already has a WhatsApp number on file. SS Care is a step up for them, not a new sale.
Live

An SS Care home. The roster changes on the day a resident is admitted or leaves — nobody types anything.

Declared

A light-path home. “2 beds free, confirmed Tue 9:40am” — true as of when the home last said so, and shown that way.

Stale

No reply for a set period. Marked unconfirmed and dropped out of matching, so nobody is sent to a bed that is not there.

Mock exchange, not built yet. Golden Years Care is fictional. This does not replace Position 4; it widens the front door so the directory is not empty while the tool is being sold.

2

A WhatsApp chatbot for families

Demand side · after the first homes are live

No app. A family messages one number, answers a few questions, and is introduced to a home that can take their parent. Runs on KengAPI, the same WhatsApp Business API platform that already sends the SS Care Flow.

The conversation, as a family would have it. Home names and budget bands are placeholders:

Family's phone · public number

Mock conversation, not built yet. Damai Care Residence, Golden Years Care and Serene Lodge are fictional homes; the budget bands are illustrative, not proposed prices.

What gets built

A matching conversation on a single public WhatsApp number:

  • Family: "looking for a nursing home for my father."
  • Bot asks, in order: area (Petaling Jaya districts to start), monthly budget range, care needs (mobility, dementia, tube feeding, diabetes monitoring, post-stroke), and urgency (this week, this month, exploring).
  • Bot matches against Phase 1 homes using their live data, and light-path homes by what they last confirmed, then returns two or three candidates with what each can take, and offers to make the introduction or book a visit. The introduction lands in the home's own WhatsApp, so the home replies from the number it already uses.
  • Where WhatsApp Flows help (a structured needs form rather than a long back-and-forth) we use them; the Flow tooling is already built and published on KengAPI.
Why WhatsApp, not an app

A family looks for a nursing home perhaps once or twice in their life, usually in a bad week. They are already on the phone with siblings, the hospital, and the homes they are considering — all in WhatsApp.

What an app asks a family to do
6 steps
  1. Stop what they are doing
  2. Find it in the store
  3. Install it
  4. Register an account
  5. Learn a new interface
  6. Never open it again
What a WhatsApp number asks
1 step
  1. Send one message

And the whole exchange sits in the same thread they will forward to their brother.

It also means zero acquisition friction for Choong: the number goes on a flyer at a hospital discharge desk, a Facebook post, a clinic notice board, and it works immediately.

Who pays

Families: free. Charging the stressed party at the point of need kills the demand side.

Homes: a referral fee per admission, or a higher subscription tier that includes enquiries fee to be decided with Choong. This is the second revenue line and the one that makes the platform, rather than the tool, the business.

The referral fee is collectable because it is verifiable: when a referred family is admitted to an SS Care home, the new resident appears in that home's roster the same day, so the platform knows without asking. Light-path homes report admissions themselves.

Data it produces

Demand by area, budget band and care need; conversion from enquiry to introduction to visit to placement; which homes respond and how fast; which needs go unmatched. That last one tells Choong which homes to recruit next, and what to build for them.

What "done" looks like

The number is public. A family can go from first message to a confirmed introduction without a human in the loop, and the home receives the introduction in its own WhatsApp with the family's needs summarised. Enquiry volume and conversion are measured from day one, because they set the trigger for Phase 3.

Risk that stops it

Too few homes to match against. If Phase 2 launches with two homes, every family gets the same two answers and the bot looks like an advert. Phase 2 should not go public until Phase 1 has enough homes across enough of Petaling Jaya to give a real choice threshold to be agreed.

3

Platform, directory, and only then an app

Only when liquidity exists

A web directory with live availability from Phase 1 data, for families who search on Google before they message anyone. The app, if it is ever built, is for homes managing enquiries, not for families searching.

What the platform layer could look like once the trigger is met. Nothing below is built, and every figure is sample data:

3aFamily-facing directory (web)Illustrative · Phase 3 · not built · sample data
homesfirst.my/petaling-jaya?care=dementia
3 homes · dementia care · Petaling JayaSorted by availability
Damai Care Residence
SS2 · Petaling Jaya
Dementia unitPost-strokeGlucose monitoringPhoto-verified records
2 beds available — updated today 3:10pm
Golden Years Care
Damansara Jaya · Petaling Jaya
Dementia careMobility supportSpO₂ monitoring
1 bed available — updated today 1:25pm
Serene Lodge
Kelana Jaya · Petaling Jaya
Dementia careTube feeding
1 bed free — confirmed by the home Fri 2:15pm · light listing

The availability line is the point. Nobody at the home edited a listing; the timestamp is the last vitals card that went to a family group. Light-path homes show what they last confirmed and when, in a different colour, so a family can tell live from declared. Damai Care Residence, Golden Years Care and Serene Lodge are fictional homes.

3b · 3cHome-facing enquiry console (app / web)Illustrative · Phase 3 · not built · sample data
Damai Care ResidenceEnquiriesHomes First console
27
residents in care · 2 beds free of 29 licensed
ReceivedFamilyCare needBudget bandStatus
Today 10:05Mdm Limvia WhatsApp · for her motherDementia careRM 3,000 – 5,000New
YesterdayMr Bongvia directory · for his fatherPost-stroke · mobilityAbove RM 5,000Visit booked · Sat 11am
2 SepMs Nurulvia WhatsApp · for her grandmotherTube feedingRM 3,000 – 5,000Visit booked · Thu 3pm
28 AugMr Wongvia WhatsApp · for his motherDementia careUnder RM 3,000Placed · Room 5 Bed B
10:05
Wednesday 9 September
Homes Firstnow
New enquiry
Dementia care · Petaling Jaya · RM 3–5k band. Reply in WhatsApp.

3c · The one place an install earns its keep: the home's admin, not the family.

This is where an app makes sense: the supply side, managing leads every day. The occupancy widget is the same count Phase 1 already keeps, so "2 beds free" is not typed in by anyone. Families never see this screen and never need an install.

What gets built
  • Web directory. Public pages per home with availability and capabilities that are true today, because they come from the same data as the chatbot. Every page has one call to action: message the number. Search-engine visibility is the reason this exists; it feeds Phase 2.
  • Home-side console. Enquiry inbox, visit scheduling, placement confirmation. This is where an app earns its place: home admins use it daily, so an install is justified. Families still never need one.
  • Choong's operator view. Homes, occupancy across the network, enquiries, conversion, revenue.
Trigger to start

Phase 3 starts only when both hold for two consecutive months numbers to be agreed with Choong:

  • N homes live on Phase 1 across enough of Petaling Jaya and its neighbouring districts that most enquiries get a real choice.
  • M enquiries a month through the Phase 2 number, with measurable introduction and placement rates.

Until then, a directory would be a page of empty listings and an app would be a store entry nobody opens.

Who pays

Homes, through the Phase 2 model, which by now has proven whether referral fees or subscription tiers work. The directory and console are what make the platform worth a higher tier to be decided.

Data it produces

Search demand by area and need (from the directory), and a full placement funnel from first search to admission. With Phase 1 data on the same residents, this becomes the only source of truth on nursing-home capacity and demand in the Klang Valley that is current rather than surveyed.

What "done" looks like

A family can find a home on the web, message it, be matched, visit, and be admitted, and the home's occupancy on the directory changes on the day of admission without anyone editing anything.

Risk that stops it

Homes unwilling to show occupancy publicly. Some homes will not want "2 beds free" on a public page. The directory can show availability as a band, or only through the bot after needs are known, with public occupancy an opt-in per home.

Decisions we need from Choong

Nothing here has a number in it yet, on purpose.

  1. Phase 1 pricing to homesSS Care monthly per home or per resident, a one-off setup fee, and the light-path fee (per day listed, or a light monthly rate). The value to the home sets the price.
  2. Phase 2 revenue modelReferral fee per admission, an enquiries tier on the subscription, or both. Families stay free either way.
  3. Phase 3 triggerThe N homes and M monthly enquiries that must hold for two months before directory or app work starts.
  4. Platform terms in the SS Care contractWhether every SS Care home is a platform participant by default, and what occupancy visibility they grant.
  5. Who sells Phase 1SoftSupplies builds and operates. Whether Choong, SoftSupplies, or both carry the home-by-home selling in the Klang Valley, and how that is shared.
  6. Data ownershipEach home owns its residents' records. Who owns the aggregated, de-identified platform data, and how it is used, should be written down before the second home signs.

What we do next, so Choong can start

An MVP is not a smaller version of Phase 3. It is the shortest path to a home that pays, a family that gets matched, and a referral fee that can be invoiced — with as little built as that takes.

The one deliberate shortcut: the family number opens with a person behind it, not a matching engine. Choong makes the first matches himself from a structured WhatsApp intake. That is not a retreat from Position 2 — families still never install anything — it just refuses to build a matcher before there is enough supply for it to have anything to choose between. The conversations it produces are what the bot is later built from.

SoftSupplies builds

Eugene · starts on go-ahead

  1. The WhatsApp botOne public number for families. Structured intake — area, care need, budget band, urgency — captured in WhatsApp and delivered to Choong as a clean enquiry. It runs on KengAPI, which is already in production.
  2. Homes First listingsThe light path: a home messages the number when a bed opens or fills, and its listing carries what was last confirmed and when. This is how the directory fills without waiting for every home to buy the tool.
  3. Choong’s dashboardEnquiries in, homes and listings managed, and the funnel measured from the first message — enquiry, introduction, visit, admission. Those are the numbers that decide the Phase 3 gate, and they are worth nothing if we start counting later.

Choong runs in parallel

Not blocked by any of the above

  1. Register the Homes First numberAnd start Meta business verification on day one. It is the slowest step in the whole plan and the one neither of us controls.
  2. Sign the first two homes on SS CareThe first proves the product. The second proves it sells twice — which is the one that matters commercially.
  3. List the light-path homesEvery home that says no to the tool is still a listing, and a listing is supply. This is how the directory fills while the tool is being sold.
  4. Answer the six decisions abovePricing and the platform-participation clause are the two that block a contract; the rest can follow.

The MVP is live when all four are true

  • Two homes record real residents on SS Care every day.
  • Light-path homes are listed and actually reply when their beds change how many, to be agreed
  • A family can message one number and be introduced to a home that can take their parent — even while a person is still making the match.
  • The first referral fee has been invoiced to a home, from an admission the platform can evidence.

What it costs Choong

Three lines. The build is a fixed price so the scope is a conversation about what gets built, not about the bill.

WhatWho paysShapeAmount
The Homes First build
The WhatsApp bot, the listings, and Choong’s dashboard — the three items in the build track above
Choong One-off, to start RM 5,000
For the scope above. Anything added beyond it is quoted separately before it is built, never after.
Running the platform
Hosting, monitoring, keeping the numbers connected, support
Choong Monthly to be decided Starts when the family number goes live, not at signing.
SS Care per home
The subscription and the one-off setup, per home onboarded
The home Monthly + one-off to be decided Billed by SoftSupplies; Choong takes a reseller margin on every home in his territory.

Then we meet again with two months of real usage and decide the matching bot on evidence: which needs went unmatched, which homes replied, and what a placement is actually worth. Phase 3 stays shut until the gate in the roadmap is met.