Agents now expose live operations APIs for analytics and intervention

Speechify Agents now expose more of the live operations layer through API-backed surfaces. Teams can stream live transcript updates, intervene in an active call, query analytics, and save shared dashboard views around their voice-agent fleet.

This release is mostly about turning monitoring from a passive page into an operating surface.

What can teams do now?

Live transcript streaming gives operators a current view of what is happening in a conversation. Take-over actions let the right person step into a call, speak, or end it when the situation needs human control.

Those actions are permission-gated independently. Listening, taking over, and ending a call are not the same permission, so a workspace can give people the minimum role they need.

Analytics and saved dashboards

The analytics API adds a typed query layer for dimensions and time windows. Saved dashboards sit on top of that so teams can build the views they actually use: call volume by agent, outcome trends, usage by workflow, or operational exceptions worth watching.

The important part is that this is backed by the control plane. The console uses the same API shapes customers can reason about, and the server owns the permission checks.

Why this matters

Voice agents are not fire-and-forget software. They talk to people. You need to know when a call is live, what is being said, when the model is stuck, and when a human should take over.

The new operations APIs make that loop shorter. Watch the call, act when needed, and inspect the aggregate behavior afterward.