Moody’s AI Skills Set to Transform Financial Analysis

Moody’s Corporation is translating centuries of analytical expertise into platform-agnostic AI skills that can be deployed through natural-language commands, marking a shift in how institutional intelligence is embedded into AI systems.

Launching initially on Microsoft 365 Copilot Cowork, the capabilities enable users to execute sophisticated analytical workflows via single natural-language requests. 

These instruction kits encode Moody’s analytical methodologies and connect AI agents to its decision-grade intelligence infrastructure. 

The software anchors all outputs in proprietary ratings, research and risk intelligence.

Cristina Pieretti, Head of Digital Content and Innovation at Moody’s, said in a company statement: “Moody’s is among the first financial data providers to deliver a full library of skills on an open standard and today’s launch is just the beginning.”

If Moody’s integrates its tools into existing financial platforms, it can become the go-to system for financial analysis industry-wide.

The approach demonstrates how domain expertise can be systematically encoded into AI agent workflows rather than requiring platform-specific integrations.

Encoding workflows into AI systems

The initial release targets high-priority financial workflows where analytical expertise is most concentrated:

• Earnings Call Summary: Processes earnings call transcripts, extracting revenue trends, pricing dynamics, consumer health indicators, tariff exposure and related metrics
• Peer Analysis: Generates comparative analysis across leverage, profitability, ESG performance, credit quality and adjacent dimensions
• Public Information Book: Assembles entity-specific dossiers spanning financials, governance structures, competitive positioning and risk profiles
• Rating Pitch: Produces structured pitch materials incorporating sector context, rating history and peer comparisons
• Sector Analysis: Synthesises proprietary research with live market data to generate sector-level intelligence.

Each skill encodes analytical procedures and quality standards designed to produce outputs that could meet consistency, sourcing and defensibility requirements for high-stakes decision-making in regulated contexts. 

A skill defines the methodology while Model Context Protocol (MCP) servers from Moody’s connect that skill to underlying data sources.

How MCP enables grounded outputs

MCP functions as an open standard that permits an AI agent to access the ratings, research and risk intelligence maintained by Moody’s directly. 

This architecture aims to ensure outputs are anchored in proprietary datasets rather than general-purpose web content, addressing a core challenge in enterprise AI deployment where hallucination and unsourced outputs remain persistent concerns.

A skill instructs an AI agent on task execution according to defined standards, with this framework captured in shareable instruction files. 

The skills are constructed using the open SKILL.md format, which originated with Anthropic and has subsequently been adopted by OpenAI, Microsoft, Google and Amazon platforms.

The open nature of the standard transforms the institutional knowledge encoded within each skill into a durable, portable asset instead of binding it to a single vendor. 

This portability means skills are built once but capable of running across any compatible platform.

Expanding the skill library

Moody’s plans to extend its skill library to encompass credit analysis, lead generation, third-party due diligence and insurance underwriting workflows.

This expansion would deploy its analytical frameworks into additional high-stakes processes where financial professionals operate.

Each subsequent skill will follow the same open, platform-agnostic standard. 

This design approach aims to ensure that institutional knowledge remains transferable across compatible AI platforms rather than requiring rebuild for each new environment.

The strategy suggests a model where domain expertise becomes infrastructure-level capability within AI systems rather than application-specific functionality.