Voice AI firm Regal launches Goal Oriented Agents with actor/director model split
Regal introduces a new voice agent architecture using two models to improve goal adherence in complex calls.
US-based enterprise voice AI company Regal Voice Inc. announced on October 27, 2026, the addition of "Goal Oriented Agents" to its platform. This feature allows businesses to define specific objectives for calls and employs two AI models working together to achieve them.
The design targets a common issue in traditional conditional-logic voice agents, such as those used for auto insurance claims: when a caller provides multiple details at once, older systems often circle back to ask for information already given just to fill each field. Regal states that building an agent on a single prompt keeps conversations open but leaves business processes loosely followed.
The new agents deploy two models per call: an "Actor," a smaller, faster model tuned for dialogue that replies immediately, and a "Director" running behind it to check if current stage goals are met and move the call to a new stage with fresh instructions. Businesses build agents from "goal nodes," listing required outcomes and prerequisites (like date of birth or mentioning a death in the family), which can be written in plain language.
According to SiliconANGLE, Regal identifies complex intake and qualification calls as strong use cases, including Medicare enrollment, roadside assistance, and mortgage qualification. The company serves over 200 enterprises, including Google LLC, AAA, and Toyota Motor Corp. Regal is a venture-backed startup that has raised $83 million to date, including a $40 million round in October 2024.