Lorikeet
for St Vincent's Health Australia
Talk to us
for St Vincent's Health Australia
for St Vincent's Health Australia
Out of every 100 patients booked for a hip replacement, maybe 15 are great candidates to recover at home — but you can't have a rich enough conversation with all 100 to find them. This demo does. A warm, meandering AI voice call pulls home setup, support network, GLP-1 status, falls history, and the unsaid anxieties — then scores candidacy and flags clinical concerns straight to your nurse pool.
Enter your mobile and the AI will ring you within seconds, as if you were Margaret O'Brien getting the pre-admission call before her hip surgery.
If Margaret were to call in to St Vincent's at any time during her treatment, the agent picks up and triages with all the same logic and integrations as your outbound assistant.
Click any prompt to copy and open chat.
Five live workflows covering the Bringing Care Home journey — pre-admission, pre-hab compliance, logistics, post-discharge check-ins, and general program questions.
Meandering 15-min conversation. Asks about home setup, support network, previous surgery, medications (esp. GLP-1), recent falls, anxieties. Scores candidacy, flags clinical concerns, books the named nurse.
What time to arrive, fasting windows, what to bring, parking, support person, medication holds. Surfaces the 7-day Ozempic washout proactively whenever a patient mentions diabetes or weight medication.
Daily SMS reminder ("done your exercises today?") → patient replies → agent acknowledges, logs compliance, or flags to nurse if the patient says they're too sore. Never says "push through pain".
Captures pain score (1-10), exercise compliance, sleep, swelling. Score ≥ 7 triggers same-shift escalation to the clinical nurse coordinator. Reads back trends so the patient sees they're improving.
Three guardrails active. Crisis indicators (chest pain, stroke signs, suicidal ideation) surface Lifeline 13 11 14 + 000. Clinical questions get deflected to the nurse pool — never answered with clinical advice. Medication questions deflect to anaesthetic team or GP.
What's Bringing Care Home? How does in-home rehab work? Costs? Eligibility? KB-grounded answers, cited from the help centre, with "I'm not a clinician" honesty when the question goes deeper.
Voice-first, greenfield-friendly. Lorikeet owns the conversation; you own the orchestration and data layer.
The agent follows the patient's lead. Six topics get covered in 4-6 turns — but in any order the patient brings up. Feels like Eileen having a chat, not a checkbox.
Every multi-fact answer is delivered in 2-3 short beats with natural pauses — not a 25-second monologue. Patients are elderly, anxious, on painkillers. They need to track + interrupt.
End of each conversation: structured candidacy score + nurse flags + recommended next steps fed back to your system. Whatever the backend ends up being — Salesforce, custom, even SharePoint — Lorikeet writes to it.
Three STEER guardrails. Crisis indicators escalate to Lifeline + 000 + nurse. Clinical questions deflect to the nurse pool with a named follow-up and SLA. Medication questions go to the anaesthetic team. The agent NEVER plays clinician.
Built around the Bringing Care Home commitment and the realities of a greenfield, no-existing-backend setup.
No existing CRM, ticketing, or clinical system to integrate with on day one. The agent's mock data layer becomes your initial system of record — and when you stand up the real backend, we just rewire the tools.
Emma AU voice at 0.9x speed for older patients. STT keywords cover Ozempic, NDIS, PRODA, Medicare, Meals on Wheels. Crisis stack is Lifeline 13 11 14 + Beyond Blue + 13YARN + 000.
Every call and chat scored against your policies (clinical safety, handover quality, nurse-flag-was-right). Patterns surface — where the agent's confident, where it's confused, where the KB needs new content.
One workflow, two channels. Voice rendering (full sentences, 3-beat cadence) and chat rendering (bullets, links) from the same source. No second "build voice later" project.
Lorikeet is built on this category long-term, with regulated-vertical customers (super, banking, healthcare) in production. You stay focused on the Bringing Care Home transformation — not on rebuilding LLM plumbing every 12 months.
An MVP version of this pre-admission flow against real Salesforce or a simple system of record is a 4-8 week build with our forward-deployed team. November is on the table.
Four steps from this demo to a live pre-admission call running against your real data.
What does a "patient record" look like in your world? We'll co-design the minimal schema — patient, surgery, home setup, support, medications. Mocks become real reads.
Run the pre-admission call against the next 100 patients in the hip queue. Score candidacy. Hand 15-20 to Eileen's nurse pool to close out. Measure agreement vs nurse-only baseline.
For the qualified candidates, turn on the SMS drip campaign (pre-hab reminders, fasting instructions, post-discharge daily check-in). Each reply pipes back through the same agent.
Hip replacement first because it's high-volume and well-protocolised. Then knee, cardiac, oncology day-care. Same agent, same Coach loop, marginal cost of each new pathway is the protocol mapping.
Happy to walk through what Phase 1 against your real systems looks like, plus connect you with the Eucalyptus team and other AU healthcare customers for reference.
Talk to the team