Voice & phone capabilities
Configure a phone agent, test the flow, and connect supported calling infrastructure.
Prepare the agent
Configure a phone agent’s responsibilities, conversation flow, runtime model, voice, and transcription settings in One. Use test conversations to inspect behavior before assigning a live phone number.
Numbers and carriers are separate resources. The available provider operations depend on the carrier adapter configured on the deployment.
Calling is an opt-in subsystem
The calling engine has separate control, public callback, and worker processes, with its own execution database, Redis, and object storage. The product API owns the user-facing configuration and access.
Only supported carrier callback and tokened media routes belong on the public listener. Internal control and engine callback routes must stay private.
Prepare for production
Enable calling deliberately with the deployment’s supported overlay. The operator must configure production policy, caller identity, allowed destinations, hours, and retention choices. Recording stays off until an approved policy permits it.
Model, speech, and carrier requests can leave the compute environment. The proposed Business plan includes voice and LLM usage with two concurrent sessions; carrier and number charges are confirmed separately.