An AI assistant for Anup's after-hours line
Takes the 2am calls so Anup doesn't have to. Triages emergencies straight to him, captures the routine stuff for the morning, and surfaces ShiftCare-shaped data for the daytime questions — km allowances, medication timing, shift handovers. Warm, calm, and clear-headed about what's clinical versus what's operational.
Try the Kinaya after-hours assistant
Open the chat below and pretend you're a casual support worker calling at 2am or mid-shift. The AI looks up real-shaped ShiftCare data, captures structured incident reports, logs sick calls — and pages Anup when (and only when) it should.
Click any prompt to copy it. The red one is the demo wow-moment.
What the assistant handles
Six action-led workflows mapped to the real call types Anup fields every day. Every conversation either resolves cleanly with the right capture, or pages Anup with full context so he picks up cold.
After-hours emergency triage
Detects participant safety / medication error / worker safety / family distress signals from the opener. Pages Anup immediately with full context, reminds the caller to ring 000 if life-threatening, and stays calm. No troubleshooting, no delays.
Sick calls + infectious flag
Identifies the worker + shift, captures the reason, flags COVID/gastro/flu so participant exposure can be tracked. Logs in ShiftCare and pages Anup with the right urgency depending on how soon the shift starts.
Incident reports (NDIS-shaped)
Structured intake: who, what, when, severity, immediate action taken. Worker's words captured verbatim, no clinical paraphrasing. Reportable incidents (death, serious injury, abuse, restrictive practice) page Anup as emergency, no exceptions.
Participant info on demand
Km allowance, default appointment length, behaviour notes, support plan basics. The bread-and-butter "I could've checked the system but I called instead" question — answered in 30 seconds straight from ShiftCare.
Medication query (safely)
Surfaces what's recorded — next dose timing, PRN eligibility with last-given time, refill status. NEVER interprets clinically. Anything ambiguous (drug interactions, dose adjustments, "should I give a half dose?") pages Anup immediately.
Casual worker onboarding
The 20-25 documents per onboarding pain. The AI tells new casuals exactly which docs are outstanding, how to send them, and when they can start — without a follow-up email chain through Anup.
Voice channel — same assistant, on a real phone line
Same workflows, same emergency triage, same warm calm tone — on voice. Emma AU voice, English STT tuned for Kinaya terminology (PRN, sertraline, restrictive practice, support plan). When we provision the phone number for your demo, Anup can call it and walk through the same scenarios live.
How we built this
Wired up around Anup's actual call types — the ones that show up most at 2am and the ones that flood his mobile during the day.
Modelled your operations
Mapped the call types Anup actually fields — sick calls, incident reports, medication queries, participant info, onboarding chase. The business context primes the AI on Kinaya's reality.
Built the workflows
Seven workflows plus a router. Emergency detection runs FIRST on every call — life/safety signals page Anup before anything else. Routine calls are handled cleanly.
Mocked your stack
Eight mock tools shaped like ShiftCare reads + writes — participant lookups, shift lookups, sick call logging, incident report capture, onboarding doc status, escalateToAnup. Swap for ShiftCare's real API in production.
Identity-free for testing
Type any name in the chat — the AI rolls with it. Lets Anup and the directors test scenarios without setting up fake worker records first.
What's next
Four steps from this demo to a real Anup-can-sleep deployment.
Wire into ShiftCare
Replace mock lookups with real ShiftCare API calls. AI answers using actual participant + worker + shift data, not approximations. Read-only at first.
Attach the phone number
Provision a phone line for the Kinaya AI to sit in front of Anup's mobile. Sit it on after-hours hours first — the highest-value time for Anup to NOT have to pick up nuisance calls.
Enable write actions one at a time
Sick call logging first. Then incident report capture. Then onboarding doc chase. Each action gated on measured success rate.
Day shift, then weekends
Once after-hours is stable, extend to the daytime info-lookup calls (km, appointment lengths, behaviour notes). And the weekends — when Anup's "off" but still on the phone.
What this means for Kinaya
Built for the reality of running a small NDIS provider single-handed.
Anup sleeps
The night calls that don't actually need him don't reach him. The ones that DO — emergencies, reportables — page him with full context so he picks up cold.
Casual workers feel heard
Anxious 2am call? Calm AI, warm tone, captures their issue properly, reassures them. Not a voicemail. Not a "we'll get back to you tomorrow."
NDIS-safe by design
No clinical advice. No medical interpretation. Reportable incidents always go to Anup. Brand guidelines hardcode the safety boundary.
AU data residency
Multi-region. Participant and worker data stays in AU. Important for NDIS privacy posture.
Phased rollout, low risk
Start read-only, enable each write action with measured success rate. The AI earns its job gradually.
Scales as Kinaya grows
Goes from 7 participants to 70 without proportional ops headcount. The agent handles the volume; the human handles the human stuff.
Ready to walk Anup through it?
Happy to do a screen-share with you and the directors, walk through what a real ShiftCare integration would look like, and answer any questions.
Talk to the team