Healthcare AI Phone Agent with Athena Integration
Patient registration and appointment scheduling by phone
We built an AI phone agent for HIPAA-compliant healthcare workflows, connecting patient calls to Athena scheduling. The build covers registration, booking, rescheduling and cancellation, with patient identity checks and confirmation of the resulting record changes.
The challenge
An appointment request needs to end with the correct change in the scheduling system. We needed the phone agent to identify the patient, find the relevant appointment and confirm the outcome before telling the caller the task was complete.
How a scheduling request moves through the agent
- 01
Understand the request
Identify the scheduling task the caller is asking for.
- 02
Resolve the right record
Establish the patient and appointment before patient-specific actions.
- 03
Use the scheduling tools
Send the intended request through the application’s Athena integration.
- 04
Verify the result
Read back the record before reporting that the change succeeded.
What we built
We connected Azure telephone and speech services to athenahealth’s scheduling system through application-controlled tools. The agent could work through patient registration, booking, rescheduling and cancellation.
Patient identity checks came before patient-specific actions. The application controlled which scheduling requests could be sent, and independent record checks confirmed the resulting changes.
Healthcare data handling
The agent was built for HIPAA-compliant healthcare workflows using Azure-hosted telephone and model services. Patient-specific access was handled through identification checks and application-controlled requests.
What we verified
Phone tests with synthetic patients exercised registration, booking, rescheduling and cancellation in Athena’s Preview environment. We checked the resulting records independently across development releases.
- These tests cover the Preview integration. Production booking volume and staff time savings have not been measured.