
September 2023 Releases
In March, we made it faster than ever to build, launch, and continuously improve AI agents with Copilot, Regal's AI agent for building AI agents. This new guided chat helps you build a production-ready agent in a fraction of the time, generating the core prompt, structure, and setup, then running tests and suggesting improvements that you can approve in minutes.
In addition, the Observability Dashboard gives you instant visibility into agent quality, surfacing latency and interruptions metrics so you can continuously monitor and fine-tune performance. Together, these updates lessen the time needed to launch and maintain agents, so you can focus on understanding your customers, unlocking new use cases, and driving faster business outcomes.
.png)
Regal is announcing Copilot, our agent for building AI agents. Most teams refine their agents over time through manual iteration and hands-on experience. Copilot lowers that learning curve by applying best practices directly into the build process, drawing on patterns developed across hundreds of millions of real customers calls. Your team simply needs to review Copilot’s suggestions and approve.
Simply describe what your agent’s objectives, workflows, and desired outcomes are, and Copilot outlines a plan and builds it, generating the core prompt, structure, and setup from a small set of inputs like scripts or transcripts to dramatically accelerate time to launch. With Copilot, you can get your first agent drafted in a few minutes, instead of hours.
To get your agent production-ready, Copilot generates call scenarios, runs tests, and suggests improvements instantly. For example, it identifies when your agent struggles to stay on topic when callers go on tangents, then propose guardrails to keep conversations focused and natural. Every suggestion comes with Copilot's reasoning, so you stay in control without getting buried in the details.
.png)
Get an instant pulse check across your AI agents’ health and quality. The Observability Dashboard surfaces call-level quality insights, including function call errors and latency metrics, so you can continuously monitor, fine-tune, and improve your AI agent. This visibility makes it easy to understand specific issues and track how your changes reduce hallucination, repetition, and robotic language, giving you the control to make every interaction feel more naturally paced and human.
Learn More
.png)
Turn every customer response into actionable data your AI agent can use to personalize the conversation in real time. Data captured during a call, or returned from an API response, can be mapped to contact fields and referenced in prompts, action parameters, and decision logic, without asking the customer to repeat themselves. For example, when a prospective student shares their major, your agent can automatically run a "Check Program Availability" action and tailor advisor recommendations on the spot, creating a seamless enrollment experience from a single response.
.png)
AI Agent Drafts let you safely build and test new iterations without risking changes to your live agents. With drafts, you can gather feedback, refine prompts and settings, and align with your team before deploying to production. Before launch, each draft version can be tested against different call scenarios in Simulations, so you can launch with confidence.
Learn More
.png)
As you add Knowledge Bases, verifying retrieval quality ensures that your AI agent will surface the right information to callers. Now you can query your Knowledge Base directly during setup and evaluate what your AI agent would retrieve, including a summary of the response, the sources it pulled from, and the specific content chunks behind the answer. For example, if you ask "Are your technicians licensed and insured?", you'll see a summary of how your agent would respond, the documents or site pages it drew from, and whether the topic is fully or only partially covered. From there, you can confirm coverage for critical questions, spot gaps, and validate new or updated material before it ever reaches a customer.
.png)
Get the context you need to validate a contact’s journey or take action with a faster, more streamlined view of their event history. Event previews on the Contact Profile surface key properties from each event directly in the timeline, so you can quickly identify call outcomes, a touchpoint’s associated campaign, and when a follow-up SMS was sent. This makes it easier to trace a contact's full journey across calls, texts, and tasks. You can also view and copy the associated task ID right from the timeline for faster investigation.
.png)
Monitor AI calls in real time and step in when it matters most. While proactive prompting can handle known escalation paths like customer frustration, the new Take Over action allows you to take control of the conversation in unforeseen scenarios or during live QA. The action seamlessly transitions the conversation from AI to a human agent without disruption, ensuring continuity and fast resolution.
Learn more
.png)
Power more precise routing and queue eligibility decisions by defining custom user attributes and assigning values to individual agents. For example, assign states, languages, support tiers, or product specializations to each agent and use those attributes directly in routing rules to ensure each call reaches the right agent.
Deploy multiple variants of an AI Agent simultaneously with different prompts, voices, or settings, and control how traffic is split between them. Whether you're testing tone, objection handling, or logic paths, variants make it easy to compare performance and double down on what works faster.
With Copilot, you can learn from real call transcripts, surface improvement opportunities, and allow teams to make changes faster, with confidence. Copilot analyzes performance signals and proposes targeted fixes, such as adding an interruption resolution rule for the claims timeline section, or refining deductible explanations. Your team can review and approve the change in one click, so your agent improves without any manual investigation.
Understand how your AI agent sounds and feels at scale with LLM quality metrics. Conversational Intelligence surfaces call-level metrics like repetition, hallucination, and tool call accuracy, so you can establish a baseline, track trends over time, and make targeted prompt or model adjustments to continuously close the gap between robotic and human.
Ready to see Regal in action?
Book a personalized demo.



