PolyAI launches Agent Development Kit for enterprise customer service
PolyAI announced the general availability of its Agent Development Kit (ADK) on April 22, a platform that lets development teams build AI agents for customer service using standard coding tools and workflows instead of proprietary interfaces.
The ADK treats agent development like traditional software engineering. Teams work in their own environment with their preferred programming language and IDE, use AI coding assistants like Cursor or Claude Code to generate logic, and manage agents through version control and code reviews.
"Most AI platforms for CX force developers to work inside a UI, cut off from the way real software gets built," said Shawn Wen, co-founder and CTO at PolyAI. "With the ADK, we're changing that. Developers can now build AI agents with the same tools, workflows, and flexibility they use to build any other critical system."
Speed gains in production
PolyAI reports significant time savings for enterprise customers. One of the world's largest utility companies built complex flows in hours instead of multiple weeks. A major UK bank processed hundreds of FAQs in minutes rather than hours of manual work.
The platform includes production patterns drawn from over 500,000 hours of real-world deployment work, allowing developers to reuse proven integrations and behaviors instead of building from scratch.
PolyAI's own engineering team uses the ADK internally. More than 60% of the company's engineering work now runs autonomously through ADK-powered workflows, with developers reviewing and guiding the output.
Continuous learning built in
The ADK includes self-learning capabilities. Agents can connect to monitoring tools and automated QA pipelines to identify performance gaps and improve without manual intervention.
This shifts agent development from static automation to systems that adapt based on real interactions.
Developer-first positioning
PolyAI positions the ADK against competitors that lock developers into no-code interfaces with limited visibility into the codebase. The company also maintains its Agent Studio platform for non-technical users, offering both code-first and no-code paths.
Enterprise customers using PolyAI include Marriott, Caesars Entertainment, PG&E, and UniCredit. The company says its largest deployments handle work equivalent to 1,000 or more full-time employees.
Developers interested in building with AI coding assistants might benefit from exploring AI Coding Courses or Generative Code Courses to deepen their skills in this area.
Your membership also unlocks:
