
September 2023 Releases
We wrapped up 2025 with releases that expand where and how you can deploy and scale your AI agents, including a new embeddable website widget for chat and WebRTC. We also released several features to increase observability across the Regal platform. Take a look at our 2025 Product Recap to revisit the highlights that pushed the platform forward this year. Thank you for your continued trust and collaboration as we head into another year of innovation.

Your website is where customers do research, compare options, and make decisions, but when questions arise, most experiences interrupt that momentum. Visitors are pushed to call and wait on hold, abandon the page to fill out a form, or interact with a generic chatbot that lacks the context, logic, or ability to take real action. That friction slows decisions, fragments conversations, and erodes trust.
With Regal’s embeddable widget, you can deploy an AI agent directly on your website using a simple JavaScript tag to handle real-time conversations across chat and WebRTC voice. Customers can start in chat for speed and convenience, escalate to WebRTC voice when nuance matters, and continue the conversation with full context preserved. Because both channels are powered by the same Regal AI agent, prompts, Knowledge Base, custom actions, and conversation state remain consistent end to end, without requiring maintenance of separate systems or logic duplication.
Embed the Regal widget on your site to
For early access to the omnichannel, embeddable widget, reach out to your Customer Success Manager.

In transfer-heavy outbound campaigns, downstream human agent availability is often the biggest constraint. With Regal’s native Five9 integration, Transfer-Aware Dialing uses synced Five9 transfer queues to pace outbound progressive dialing in Regal based on both the target campaign queue and downstream transfer capacity. This ensures AI Agents can qualify leads at scale without overwhelming licensed or specialized Five9 agents downstream. By aligning dial volume to real-time transfer availability, you can reduce abandoned handoffs while keeping wait times low and maximizing the ROI of every qualified conversation.
For early access to Transfer-Aware Dialing, reach out to your Customer Success Manager.

AI Agents can now reliably collect numeric information through caller keypresses, giving you a dependable alternative when spoken input is slow, error-prone, or sensitive. Keypad input is ideal for capturing account numbers, PINs, ZIP codes, menu selections, or other identifiers that the agent can reference later in the call or pass to downstream systems. You control how inputs are collected, with support for configurable timeouts, termination keys, and digit limits that align to your use case. All captured keypad inputs appear in the post-call transcript within the `digits_pressed` action invocation, allowing you to validate exactly when and how the input was collected.
For early access to Keypad Input Detection, reach out to your Customer Success Manager.

You can now create test cases directly from call transcripts with a single action, turning real production conversations into repeatable simulations. Generate tests to validate how your AI agent behaves when things go wrong in production, such as missed actions, incorrect transitioning, or inaccurate Knowledge Base retrievals. By turning real customer interactions into test cases, you can expand coverage, validate agent behavior after prompt or Knowledge Base changes, and catch regressions earlier using scenarios that reflect real-world performance.
.png)
Call transcripts now surface state transitions and action errors directly alongside the conversation, providing execution-level visibility into agent behavior and where it breaks down. For multi-state agents, each transition is shown inline so you can trace how the agent moved between nodes, such as entering an escalation state too early, skipping a required qualification step, or falling into an unexpected fallback, and verify that flows are executing as designed. Action errors are aggregated in the Call Overview section and link directly to the underlying action configuration, allowing you to pinpoint issues like invalid transfer destinations or failed API calls and quickly remediate them.

You can now view a queue’s qualifying agents directly alongside its definition in Settings, with live visibility into the total number of agents currently eligible to receive tasks and each agent’s real-time status. Validate queue definition changes on the spot or diagnose why a specific agent isn’t receiving the tasks you expect.

You can now hide specific queues from human agents’ internal transfer dropdown, ensuring agents only see valid transfer destinations. Exclude system queues, AI-only queues, or restricted escalation queues that frontline agents should not access. Limiting transfer visibility reduces accidental mis-transfers, protects sensitive workflows, and keeps internal handoffs aligned with your operational design.
.png)
Custom views can now be saved on the Journeys, Campaigns, and Segments pages. Save views for active launches, underperforming campaigns, priority segments, or in-flight journeys and return to them instantly without reapplying filters. By reducing repetitive setup, saved views support faster collaboration and day-to-day execution across your team.
Manage Teams directly in-app by updating names and descriptions or deleting them entirely.
Automatically gather critical information during live calls using Regal’s AI Agent, then reference it later in prompts or actions in a more structured way to keep the conversation moving without repeated questions.
Bring customer context into the call before it begins using GET requests, enabling smarter IVR routing and more personalized agent handoffs.
Ready to see Regal in action?
Book a personalized demo.



