Duck Creek Releases Agentic AI Platform for Insurance Operations
Duck Creek Technologies has introduced an AI platform designed to let insurers build and manage AI agents that work across underwriting and claims workflows. The platform combines data from insurance core systems with machine learning and rule-based reasoning to keep agents operating within established processes and compliance frameworks.
The company released two initial applications built on the platform: one for underwriting submissions and another for first notice of loss (FNOL) processing. Both aim to reduce manual work and processing time in workflows that typically involve multiple handoffs and data entry steps.
How the platform works
The Agentic AI Platform operates across five layers. The intelligence layer uses insurance-specific models, fine-tuned generative AI, and neuro-symbolic reasoning-a combination of deterministic rules and probabilistic learning-to give agents domain knowledge. The orchestration layer provides a central environment for creating and coordinating agents across different use cases while maintaining human oversight.
An assurance layer handles governance: traceability, audit records, monitoring, and compliance controls so that AI decisions can be explained. An integration layer supports standard protocols and connects to Duck Creek's core systems and external platforms. The foundation layer accesses live policy, claims, billing, and risk data directly.
Duck Creek designed the platform to work with both embedded and headless deployments, meaning insurers can integrate AI capabilities into existing systems without replacing them entirely.
The two initial applications
The Agentic Underwriting Workbench receives submissions, assesses and enriches them in real time, and produces decision-ready files. The company says this reduces turnaround times and improves risk assessment by automating data collection and prioritizing high-value opportunities.
The Agentic FNOL application coordinates agents across digital, voice, and mobile channels to capture, check, and route claims immediately upon intake. Built on Google Cloud and powered by Gemini models, it includes early-stage policy validation and fraud detection at the point of claim submission.
Broader strategy
Duck Creek frames these applications as examples of a shift from isolated task automation toward coordinated, end-to-end workflow execution. Hardeep Gulati, the company's CEO, said the platform combines "core systems, proprietary insurance domain ontology and expertise, and Agentic AI with neuro-symbolic reasoning to create agents that operate with full context, governance, traceability and human in the loop."
The approach reflects how insurers are adopting AI for Insurance operations-moving beyond single-task automation to coordinated decision-making across multiple personas and stages of the customer lifecycle. Learn more about AI Agents & Automation in professional settings.
Your membership also unlocks:
