Regal Adds Two-Model AI Agents to Guide Complex Business Calls

Regal introduced Goal Oriented Agents, a voice AI feature that lets businesses define call outcomes and track whether required steps are complete. The system is designed for conversations in which callers may provide information in a different order than a fixed script anticipates.

Each call uses two models in parallel. A smaller, faster actor model handles the spoken dialogue. A director model checks the active stage’s goals and prerequisites after each caller turn. It can guide the actor’s next response or move the call to another stage.

Businesses define requirements in goal nodes. A node can specify information to collect or an action to take, such as scheduling a callback. Requirements can be structured data fields, such as a caller’s date of birth, or conditions written in plain language. Businesses can use these nodes to set what a call needs to accomplish without scripting every question in a fixed order.

The two-model design separates conversation from progress tracking. The actor responds to the caller, while the director checks which requirements for the current stage have been met and determines whether to guide the next response or advance the call. The setup aims to keep a conversation flexible while tracking required business steps.

An illustrative auto insurance example shows a caller sharing details about vehicle damage, photo evidence and injuries in one turn. The system marks those items as collected and continues by asking for information that is still missing. The example does not describe a confirmed customer deployment.

Potential applications include complex intake and qualification for auto insurance claims, Medicare enrollment, roadside assistance and mortgage qualification. Medical intake involving HIPAA compliance is another potential use. These are applications for the feature, not confirmed customer deployments.

Existing Regal customers seeking access can contact their customer success manager.