Case study — Sophie Platform

Converting missed calls into structured, auditable interactions

Uniting AgeWell operates after-hours care services across Victoria. One in five calls was going unanswered. Taidotech deployed Sophie as a governed safety net — converting every missed call into a structured, triage-classified record.

Organisation
Uniting AgeWell
Aged care provider — home care, residential, and retirement living services across Victoria
Challenge
18.5% unanswered call rate
3,604 of 19,456 after-hours calls going to voicemail with no structured capture
Solution
Sophie — Growth tier
Managed voice + actions platform deployed as after-hours safety net, integrated with AlayaCare
Regulatory context
Aged Care Act 2024
Strengthened Quality Standards, SIRS obligations, and continuity of care requirements

The challenge

Between January and August 2025, Uniting AgeWell's after-hours service received 19,456 calls. Of these, 3,604 — approximately 18.5% — went unanswered and were diverted to voicemail or alternative channels. There was no structured record of who called, why they called, or what happened next. Follow-up depended on manual voicemail review, which was uneven and time-consuming.

The consequences were familiar but previously unquantified: client experience degraded by reaching voicemail at a moment of need; clinical and welfare signals potentially missed or delayed; no reliable audit trail of after-hours demand; and the after-hours team's capacity fully consumed by the calls they could answer, with no mechanism for the ones they couldn't.

Under the Aged Care Act 2024 and the Strengthened Quality Standards, the governance expectations around continuity of care, responsiveness, and documented client interactions had increased materially. The existing approach — manual voicemail review, ad-hoc follow-up — was increasingly difficult to defend in an audit or Commission review context.

The governance gap

"The after-hours team was doing exactly what the current system asked of them. The problem was the system had no mechanism for calls that arrived when the team was on another call or unavailable. Roughly one in five callers was receiving a voicemail rather than a response — and there was no structured record of any of those interactions."

The approach

Taidotech conducted structured workshops with Uniting AgeWell in late 2025, mapping the after-hours operating model, the primary call types, the alerting and escalation requirements, and the governance boundaries that applied in an aged care context. This co-design process identified the main after-hours call patterns, the SIRS-related escalation obligations, and the integration requirements for AlayaCare.

Four call patterns were then validated end-to-end in Taidotech's sandbox environment against a representative mock AlayaCare database — covering home care worker shift cancellations and reschedules, client welfare concerns, family and next-of-kin enquiries, and general service enquiries. The POC phase tested caller identification, conversational quality, alert routing through Microsoft Teams, and reporting outputs to Excel and SharePoint.

The deployment model was deliberately staged. Sophie was configured as a safety net — activating only when a call was not answered within the agreed threshold. The after-hours team's workflow was unchanged. Sophie sat behind the existing service, handling only the calls the team couldn't reach.

What was deployed

Sophie was deployed on the Growth tier, integrated with AlayaCare for real-time caller identification and context retrieval. The platform was configured for four validated call type patterns covering the highest-volume after-hours scenarios.

  • AlayaCare integration — read-only API connection enabling phone-number-based caller identification and client context retrieval before the conversation begins, linking every interaction to the correct client record where a match exists.
  • Deterministic triage framework — urgency classification rules configured jointly with Uniting AgeWell's clinical governance team, covering routine, elevated, and urgent alert levels including SIRS-related indicators. No unsupervised AI classification.
  • Microsoft Teams alerting — real-time alerts routed to the after-hours team channel with full interaction context, triggered for elevated and urgent classifications.
  • Sophie Insights Centre — full operational dashboard, conversation review workspace, and caller journey analytics. Every interaction available for review with transcript, audio, structured summary, and triage classification.
  • Excel and SharePoint reporting — scheduled exports of call logs and summaries for operational review and audit purposes.

POC results

The following results are from Taidotech's controlled POC phase against the four validated call patterns, using a representative mock dataset. Live operational metrics are tracked continuously through the Sophie Insights Centre.

POC results

Measured in controlled testing against four validated call patterns.

100%
Call containment
Every test call handled to completion by Sophie without human handover. Directly measured in POC.
95%
Task success rate
Interactions where required information was captured accurately against the target pattern. Directly measured.
89%
Alert accuracy
Alerts triggered correctly against configured rules under conservative settings. No unintended triggers observed.
3:55
Avg. call duration
Consistent with the 3–5 minute planning assumption used in commercial scoping. Directly measured.

These values are from controlled POC conditions with a representative mock dataset. Live operational metrics are tracked through the Sophie Insights Centre and reviewed with Uniting AgeWell on a periodic cadence.

The governance outcome

The deployment replaces voicemail with a governed interaction layer. Every call that would previously have ended in voicemail now generates a transcript, a structured summary, a triage classification, and — where warranted — a real-time alert to the after-hours team through Microsoft Teams. The governance position is materially stronger: where previously there was no record of who called or why, there is now a complete, searchable, exportable audit trail of every interaction.

The clinical and welfare signal that might previously have been lost in a voicemail inbox is now captured, classified, and — if it meets the configured urgency threshold — in front of the right person within seconds. The after-hours team's workflow is unchanged. What changed is what happens to the calls they can't reach.

"The pilot replaces the voicemail experience with something materially better, without changing how the after-hours team operates. If Sophie is switched off, the service reverts to its current state. The risk surface is small and the exit path is clean."

Technology deployed

Voice platform
Sophie — Growth tier
Managed voice + actions platform. Inbound call handling, natural conversation, structured capture, triage, and alerting.
AI infrastructure
Azure AI Speech + Foundry
Real-time speech recognition and synthesis via Azure Communication Services and Azure Speech Services. Hosted in Azure East Australia.
Care management integration
AlayaCare API
Read-only integration for real-time caller identification and client context retrieval. Phone-number-based matching as the primary identification mechanism.
Alerting
Microsoft Teams
Real-time alerts with full interaction context routed to the after-hours team channel. Configurable by urgency level and routing destination.
Reporting
SharePoint + Excel
Scheduled reporting exports of call logs, summaries, and triage classifications for operational review and audit purposes.
Governance
Sophie Insights Centre
Operations dashboard, conversation review workspace, caller journey analytics, full audit logs, and role-based access control.