Seeking social housing partners

Ruby

The check-in call that keeps a tenant from going unseen.

A warm, scheduled voice call for social housing tenants — staying alongside people between visits, and quietly flagging anything a housing team should follow up. Every concern is reviewed by people. Ruby supports housing providers and welfare staff. It never replaces them.

The gap between visits

Around six in ten social housing households are a single adult living alone1 — and the contact that should reach them often doesn't.

Australia's main social housing programs support roughly 830,000 people, and the sector is old and ageing — two-thirds of main tenants are aged 50 or over, and the 75-plus group is the single largest.2 Many live alone, with a stable tenancy but thin day-to-day contact. It is the structurally highest-risk profile for a quiet crisis to go unnoticed.

The contact obligations exist — but they are hard to meet at scale. In NSW, only 58% of new social housing tenancies received the required welfare visit within 12 weeks, against a 95% target (year to June 2024).3 And tenants do go unseen: in the UK, one tenant lay dead for around two and a half years despite 89 contact attempts — the inquest found policies were "followed", and the failure was siloed, task-focused work, not absent policy.4 Australia has seen comparable cases.5 That gap — consistent, person-centred contact that someone actually completes and records — is what Ruby exists to close.

~6 in 10

social housing households are a single adult living alone1

2 in 3

main tenants are aged 50 or over2

58% vs 95%

new NSW tenancies got the required welfare visit in time, against target3

43%

of social housing tenants live with a mental health condition — highest of any tenure6

How it works

A warm conversation that quietly does careful work.

A gentle, scheduled call

Ruby phones tenants on a cadence the provider sets, for an unhurried, human-feeling chat — wellbeing, home and repairs, social connection, "anything you need a hand with". No form to fill in. The cadence is configurable per provider policy, never marketed as clinically derived.

Modelled on established outreach

Ruby's call-and-escalation pattern is modelled on long-running telephone-outreach services — like Red Cross Telecross, a free daily wellbeing call to thousands of at-risk Australians with a defined no-answer escalation chain.7 Ruby brings that proven human model to more tenants, consistently.

Flags for housing-team review

When Ruby hears something that warrants a closer look — arrears or disrepair distress, isolation, health deterioration — it surfaces a clear, plain-language summary for a staff member to review and decide on next steps. People, not the AI, make the calls.

Engineered escalation — humans in the loop

A tenant who can't be reached triggers a hard-coded ladder — retry, SMS, nominated contact, a task to the tenancy manager, and ultimately a provider-initiated welfare check per its own policy — every step timestamped for audit. This path is built, tested and rehearsed, not left to chance.

The evidence, briefly

Designed from the research, honest about what's proven.

Why scheduled calls

Scheduled welfare calls to isolated people are an established, trusted Australian model: Red Cross Telecross delivers a free call 365 days a year to 5,000+ at-risk Australians, with an engineered escalation chain.7 Ruby's cadence is modelled on that human outreach. Ruby itself is a welfare support-and-connection tool — not a treatment — and we treat evaluation as ongoing.

Why it helps sustain tenancies

Human-delivered sustaining-tenancy programs hold 80.9–92.3% of at-risk tenancies, with early intervention and outreach the named success factors8 — while homelessness costs governments an estimated A$29,450 per person per year in extra services.9 Ruby is an early-flag layer feeding those referral pathways, not the support program itself.

Why the gap is real

DCJ policy already requires periodic welfare contact with older tenants — yet completion runs well below target.3 Ruby maps directly onto an existing policy obligation, and its call data is reportable evidence for NRSCH tenant-engagement performance outcomes10 and CHIA ESG social-outcome reporting.11 Contact requirements exist; capacity to complete them doesn't. Ruby extends reach without pretending to be the welfare check itself.

Who Ruby is for

Built to sit alongside the people already supporting tenants.

State housing authorities

Meet periodic tenant-contact obligations consistently, with every contact documented for audit.

Community housing providers

Evidence NRSCH tenant-engagement outcomes and ESG social metrics with structured call data.

Tenancy & welfare teams

Add a light-touch check-in between visits, with clear summaries when something needs attention.

Sustaining-tenancy programs

Surface emerging risk early so referral pathways like Together Home and Tenancy Plus engage sooner.

Powered by Kate

Every call Ruby makes is orchestrated by Kate, the coordination engine behind all CAREPLANS AI companions. Kate manages scheduling, conversation, risk flagging and the engineered escalation path across every persona and every vertical.

Safety, privacy & what Ruby is not

Careful by design, for a population where care matters most.

A welfare support tool — not clinical care

Ruby supports tenant welfare monitoring and sustaining tenancies. It does not diagnose, treat or provide clinical or mental-health care, and is not a medical device. It checks in, listens and flags for a staff member to review. Clinical and tenancy decisions always rest with qualified people.

Genuine distress goes to human help — by design

Ruby is not a crisis service and does not replace statutory or police welfare checks. Independent testing has shown generic AI chatbots routinely mishandle moments of crisis12 — which is exactly why Ruby's escalation path is engineered and rehearsed: urgent concerns are priority-flagged to a human, and a tenant in distress is pointed to Lifeline 13 11 14, 13YARN 13 92 76, or 000 in an emergency.

Data handling

Data is stored in AWS Sydney (Australia). AI processing currently runs in the United States (Anthropic and Hume), with zero-data-retention in progress. We never train on customer data.

Security posture

Essential Eight Maturity Level 3 controls implemented; ISO 27001:2022 aligned, with certification in progress. Ruby is not a certified medical or clinical device. Built on Claude and Hume EVI.

If you or someone you know needs support now: Lifeline 13 11 14 · 13YARN 13 92 76 · Beyond Blue 1300 22 4636 · Red Cross Telecross 1300 885 698 · In an emergency call 000.

Pilot · in discussion

Let's make sure no tenant goes unseen.

Ruby is in pilot, and we're in discussion with state housing authorities, community housing providers and welfare teams who want to stay alongside tenants between visits. If that's you, we'd welcome the conversation.

andrew@careplans.io

Sources

  1. Australian Institute of Health and Welfare, Housing assistance in Australia (2024): ~58% of public housing and ~63% of community housing households are a single adult living alone. Single-adult shares via AIHW summaries; pin exact table/year before formal use.
  2. AIHW, Housing assistance in Australia (2024): ~830,000 people in main social housing programs; two-thirds of main tenants aged 50+, with the 75+ group the largest single main-tenant age band.
  3. Audit Office of NSW, Social housing reporting (year to June 2024): only 58% of new DCJ/Homes NSW tenancies received a successful welfare visit within 12 weeks, against a 95% target. Contact-frequency obligations: NSW DCJ Tenancy Policy Supplement.
  4. Sheila Seleoane / Peabody inquest (July 2022): tenant found dead ~2.5 years after death despite 89 contact attempts; independent Altair review attributed the failure to large warden patch sizes and siloed, task-focused culture — "the customer got lost". Inside Housing, 2022.
  5. NSW Coroner, Inquest into the death of Atilla Demirer (finding 2023): Waterloo public-housing tenant undiscovered for up to ~3 months; cause/manner undetermined. Cited as illustration of an Australian failure mode, not as a death-rate statistic — no systematic Australian registry of "deaths unnoticed" exists.
  6. Peer-reviewed analysis of Australian survey data (2023): ~43% mental-health-condition prevalence among social housing tenants — the highest of any tenure. Single 2023 analysis; cross-check against AIHW NSHS 2023 before headline use.
  7. Australian Red Cross Telecross: a free daily wellbeing call, 365 days a year, to 5,000+ at-risk Australians, with a defined no-answer escalation chain (retry → nominated contact → police welfare check). Decades of operation; an established service model rather than an RCT-proven intervention.
  8. Australian Housing and Urban Research Institute (AHURI), What is a sustaining tenancies program? and Final Report 252 (Zaretzky & Flatau 2015): sustaining-tenancy programs hold 80.9–92.3% of supported tenancies (evictions 0.3–3.4%), with early intervention and outreach as named success factors.
  9. AHURI (FR218/FR252, 2015–16-era modelling): homelessness costs governments an estimated ~A$29,450 per person per year in additional services. Quoted with vintage.
  10. National Regulatory System for Community Housing (NRSCH), National Regulatory Code: "tenant and housing services" is a mandatory performance outcome requiring demonstrated tenant engagement, fair treatment and satisfaction evidence.
  11. Community Housing Industry Association, ESG Reporting Standard (launched March 2023; second annual review FY23–24): voluntary sector standard with tenant-wellbeing and satisfaction social-outcome criteria, designed to support ESG-screened institutional lending.
  12. 2025 evaluation (reported 2026) of 29 mental-health and general chatbots found none responded adequately to suicidal messaging, with wrong-country crisis numbers common (Pichowicz et al. 2025; Scienceline, April 2026). Ruby's design assumes AI alone is not safe for crisis moments.

Statistics describe population and sector research, not Ruby's own outcomes; Ruby's effectiveness is under evaluation. Crisis and service phone numbers are verified to the cited sources at time of writing and re-checked before any tenant-facing deployment.