Decagon Launches Self-Improving AI Agent for CX

Decagon helps enterprises improve AI customer service performance by equipping agents with automated learning, testing, and optimisation.

Decagon has introduced Duet Autopilot, a self-improving AI agent designed to help organisations continuously optimise customer service performance. The company said the platform automatically identifies improvement opportunities, tests changes, and recommends updates while maintaining human oversight and governance.

According to Decagon, traditional AI agent optimisation often relies on manual analysis, testing, and deployment cycles. Duet Autopilot is designed to automate much of that process by continuously analysing production interactions and generating improvements based on customer conversations and operational signals.

“Autopilot is a shift from building agents by hand to managing agents that improve themselves. Teams set the direction and review the work; Autopilot handles the diagnosing, testing, and editing that used to consume their week. Every fix compounds, which ultimately empowers businesses to provide their customers with a 24/7 AI concierge that gets measurably better with every interaction,” said Alan Yiu, VP of Product, Decagon.

ALSO READ: Human Agents Pay the Hidden Operational Cost of AI in Customer Service

Automating Continuous AI Improvement

The platform operates through three core capabilities: automated improvement, self-validation, and enterprise governance.

Decagon said Autopilot continuously identifies opportunities to improve customer interactions, creates proposed updates, and validates those changes against historical conversations, regression tests, and predefined customer scenarios before presenting them for approval.

The company noted that every update remains subject to human review, with teams able to define brand guidelines, policy requirements, writing standards, and operational guardrails that the AI must follow.

Supporting Better Customer Outcomes

Duet Autopilot is currently being evaluated by enterprise organisations across financial services, retail, and consumer technology sectors. According to Decagon, customers are using the platform to improve resolution rates, reduce escalations, and expand automation coverage across customer service operations.

“At our scale, manually reviewing conversations for errors isn’t an option. Decagon Autopilot frees our team to focus on decisions rather than digging through logs. It surfaces what changed, what was considered, and why. That transparency is what makes AI actually trustworthy in production,” said Matt McCollum, Senior Manager of Customer Experience, Opendoor.

Alongside the launch, Decagon introduced DuetBench, a benchmark designed to evaluate AI agent self-improvement capabilities. The company said the framework measures whether AI agents can generate and validate meaningful performance improvements rather than simply produce conversational responses.

The launch reflects growing interest in autonomous AI systems that can continuously optimise customer experiences while maintaining transparency, governance, and human control over service quality.

ALSO READ: Turning Feedback into a Strategic Brand Signal

- Advertisement -spot_img

Featured Articles