How Sophie works
Sophie combines four layers into a single managed service.
Sophie answers inbound calls and can make outbound calls. She uses natural Australian English, handles interruptions, adjusts her pacing to the caller, and maintains a conversational rhythm that feels human. Behind the scenes: Azure Communication Services for call handling and Azure Speech for real-time speech recognition and synthesis.
Sophie doesn't just listen and respond. She follows governed workflows: identifying the caller, capturing the reason for their call, classifying urgency, triggering alerts, and routing information to the right systems. Every workflow is configured, tested, and controlled — Sophie does not improvise.
Sophie connects to your systems. She can look up a caller in your care management system, confirm a roster in SharePoint, send an alert to Microsoft Teams, create a structured record, or export a report. These are governed actions, not open-ended API calls. Sophie does what she's configured to do, nothing more.
Everything Sophie does is visible. The Insights Centre is a web-based dashboard where your team can see call volumes, review conversations, inspect how Sophie handled each interaction, manage alerts, and track operational performance. It's the governance surface that makes Sophie auditable and reviewable.
Caller journey
Live, governed, and visible end-to-end. Switch scenarios to see how the same platform handles different conversations.
Insights centre — live capture
Inbound call arrives at the after-hours number.
How a call works
A call arrives at your number. Depending on your configuration, Sophie may answer directly, handle overflow when your team is at capacity, or activate when a call isn't answered within an agreed threshold.
Sophie answers immediately and begins a natural conversation.
Where possible, Sophie identifies the caller against your source system — by phone number, name, or other configured method.
Sophie captures the reason for the call in a structured format, following the configured conversation flow for that use case.
Sophie classifies the interaction by urgency using deterministic rules — routine, elevated, or requiring immediate escalation.
If an alert is triggered, it's sent to the configured channel (Teams, email) in real time.
If downstream actions are required — a system lookup, a record update, a report — Sophie executes them within the governed workflow.
The full interaction — recording, transcript, structured summary, and triage classification — is logged in the Insights Centre.
Your team can review, act, and follow up with full context.
The Insights Centre
Total call volumes, outcomes, alert counts, demand patterns by time of day and day of week. For the first time, you'll have reliable data on what your service is actually being asked for.
Every interaction is available in a split-panel workspace: full transcript alongside the original audio. Jump to specific conversation stages, inspect how Sophie interpreted the caller's intent, and confirm that the summary matches what was said.
Interactions presented as a flow — showing how callers move from entry through conversation stages to outcomes. This exposes patterns that are invisible in summary statistics.
Full audit logs, role-based access control, change history, and retention management. Every action Sophie takes is traceable. Every configuration change is versioned and reversible.
Real-time alerts to Microsoft Teams or email when trigger conditions are met. Configurable by urgency level and routing destination.