The architecture of Voice AI consists of the Voice AI desktop app, multiple cloud-based services, and the Pega Customer Service application. Between them, these components capture, route, and process voice interactions before displaying the real-time suggestions to CSRs.
- The Voice AI desktop app captures audio streams from softphones on agent computers and routes them to other components for further processing.
- Cloud-based multi-tenant services convert speech to text, manage security, and train audio models. Pega hosts these services and automatically makes them available to all Voice AI clients.
- The Pega Customer Service application runs transcripts through Pega Platform natural language processing (NLP) and pushes suggestions to agents in real time in the Interaction Portal.
The following figure illustrates how various components in the Voice AI architecture work together to deliver suggestions to CSRs: