Introducing CatMoS.AI-SlipStream: Agentic AI for intelligent policy slip interpretation

Authors: Tathagata Malakar - Senior Director, Moody's; Satyanarayana Guda - Director, Moody's

To empower insurers with increased, unparalleled efficiency and consistency in managing their catastrophe exposure, CatMoS.AI, Moody's new leading-edge catastrophe modeling workflow solution, leverages the power of agentic AI to streamline complex workflows, enhance decision-making, and ensure robust risk insights across the insurance value chain.

A key module within CatMoS.AI is SlipStream, which accelerates processes and optimizes workflows, enabling insurers to stay ahead in a rapidly evolving risk landscape.

The significance of the insurance slip

The ‘slip’ is a time-honored, foundational insurance record. Its origins can be traced back some 600 years; a document that contains all the pertinent information regarding the risk and the insurance terms and conditions.

Across global insurance markets, brokers submit slips to underwriters; if an underwriter decides to participate on the risk, they record their percentage and pricing and sign the slip. The process is then repeated until the slip is filled.

Slips are important; they form the basis for the insurance coverage contract. For example, in the event of a difference in wording between the slip and the policy issued from it, the slip supersedes the policy as the binding insurance document.

Over many years, Moody’s Analytical Services has amassed deep expertise in helping clients process data, totaling over 60,000 insureds across global markets and managing approximately 25,000 policy slips annually.

We understand how policy slip interpretation is foundational for catastrophe risk assessment. Accurate extraction of financial terms—limits, deductibles, participation structures, and coverage conditions—determines the quality of downstream exposure modeling and underwriting decisions.

Challenges of slip processing

Each slip presents distinct challenges: formats ranging from scanned images to digital PDFs, document lengths spanning 20 to 200 pages, and critical financial conditions embedded within complex contractual language. Manual processing typically requires 30 to 90 minutes per slip, depending on complexity.

This creates persistent resource constraints, limiting organizational capacity to scale operations or respond to peak demand periods. The process has historically resisted automation due to the inherent variability in insurance documentation and the domain expertise required for accurate interpretation. Can GenAI help to accelerate slip processing, without compromising data quality?

Moody's vision for the future state of account modeling is a fully orchestrated workflow where specialized AI agents collaborate with human experts across the entire processing pipeline.

This ‘Digital Co-worker’ environment represents a fundamental transformation in how catastrophe risk analytics are produced, and we are delighted to introduce SlipStream, an agentic AI application that automates the interpretation of insurance policy slips for catastrophe modeling workflows.

What is SlipStream?

The SlipStream agentic AI application transforms unstructured policy documents into structured, analytics-ready data—delivering the interpretive accuracy of experienced analysts at enterprise scale, focused around three core principles:

Context-aware interpretation: SlipStream leverages an advanced agentic AI framework to blend decades of insurance domain expertise with real-world variation—unusual document structures, regional conventions, client-specific terminology, and contractual edge cases. The system delivers context-aware interpretation, not just rote extraction, to understand how financial terms function within broader policy structures.

Digital co-worker model: Central to SlipStream's architecture is the coworking interface, where analysts and AI work side by side to validate and refine outputs in real-time. Every model correction and revision is reviewed, debated, and improved before being incorporated into the application’s learning pipeline. As processing volumes increase, so does system intelligence—the more accounts SlipStream processes, the smarter it becomes.

Dynamic learning architecture: Unlike static automation systems that apply fixed rules regardless of context, SlipStream continuously absorbs expert feedback, transforming every account's nuance into teachable intelligence. This dynamic feedback loop—where disagreements become data and collective experience becomes logic—drives accuracy and resilience even under operational pressure.

SlipStream helps add value by increasing processing efficiency, accuracy, and deepening expertise: 

How SlipStream works

SlipStream operates as a digital coworker within existing analyst workflows. The system ingests policy documents in any common format—scanned images, PDFs, Word documents, or plain text—and processes them through a series of context-aware AI agents. Each AI agent is specialized for specific extraction tasks: identifying document structure, locating financial terms, interpreting conditional language, and mapping extracted data to standardized output templates.

The underlying AI framework combines large language model (LLM) capabilities with domain-specific training derived from decades of Moody's insurance expertise. This hybrid architecture enables the system to handle the terminological variation and structural inconsistency that characterizes real-world policy documentation.

Analysts interact with SlipStream through a collaborative interface that surfaces AI-generated interpretations for validation, correction, and refinement. Expert feedback is systematically incorporated into the learning pipeline, creating a continuous improvement cycle where the system becomes more accurate and more attuned to industrial interpretation standards over time.

Market validation

Completing extensive client evaluation cycles, SlipStream has consistently gained a positive reception. Live demonstrations have generated strong interest from organizations managing high-volume policy portfolios, with several clients expressing intent to pursue licensing arrangements.

One telling indicator is that organizations with internal GenAI development programs have noted that SlipStream's capabilities exceed their current initiatives—a reflection of the specialized domain training and production-grade architecture underlying the system.

"While we are working on developing GenAI applications internally, none can compare to what Moody's presented to us."

— Client feedback, product demonstration

Future-ready architecture: The digital coworker vision

SlipStream represents the foundational step toward realizing this ambition. As the Slip Agent within the Digital Coworker ecosystem illustrated above, SlipStream has been architected not as a standalone solution but as an interoperable component designed to operate alongside adjacent agents.

Within the Moody’s environment, SlipStream runs in parallel with the Data Cleansing Agent that handles automated data cleansing, validation, and preparation of schedules of values. Both systems contribute their respective outputs to a unified account modeling process.

This parallel architecture ensures that policy interpretation and exposure data preparation proceed simultaneously, converging into a consolidated output that feeds downstream analytics.

Similar CatMos.AI modules for exposure enrichment, portfolio rollup, pricing, and reporting will soon be introduced.

Get Started: To schedule a demonstration or discuss implementation requirements, contact your Moody's Analytics representative.