Best use of AI: Moody’s

Moody’s Banking Decision Intelligence solution uses artificial intelligence to bring together credit, liquidity, capital, profitability and other insights, helping banks make faster decisions without sacrificing transparency, governance or regulatory defensibility

Banks are awash with data from all sorts of different systems. Credit risk and underwriting, profitability models, capital planning tools, liquidity platforms and portfolio analytics engines generate vast amounts of information every day. Yet many important decisions still depend on analysts manually pulling together insights from across these systems before arriving at a conclusion.

Moody’s believes that process has become unsustainable.

Moody’s Banking Decision Intelligence solution, winner of the 2026 Risk Technology Award for Best use of AI, was developed to address a challenge facing many financial institutions: how to synthesise fragmented data, models and analytics into a coherent decision-making framework without compromising governance or explainability.

Instead of introducing another dashboard or point solution, Moody’s Banking Decision Intelligence acts as an enterprise-wide intelligence layer that sits across a bank’s existing infrastructure. It connects lending, credit, current expected credit loss (CECL), asset-liability management, liquidity, capital, profitability and portfolio management systems, enabling users to ask business questions in natural language and receive a thoughtful synthesis, supported by transparent analytics and auditable calculations.

Banking decisions rarely sit neatly within a single function. A relationship manager considering whether to reprice a loan may need to understand not only the profitability implications, but also the impact on capital consumption, funding costs and portfolio risk.

Similarly, a treasury team monitoring liquidity exposure may benefit from understanding how those risks connect to lending activity, deposit behaviour or broader balance sheet dynamics. 

The solution integrates a bank’s internal data with Moody’s analytical engines and models, assembling the relevant information automatically before synthesising contextualised responses with options and analysis on their trade-offs. Every synthesis is accompanied by supporting assumptions, calculations and data lineage, providing transparency for both internal governance and regulatory requirements.

That synthesised output still has to be interrogated before anyone acts on it. “You need to triangulate that data within the context of other data,” says Charlene Bian, Moody’s head of risk and finance solutions. “You need to benchmark it, you need to compare it, you need to be able to fully wrap your head around it before you can act on it confidently.”

While many organisations are exploring AI-enabled decision support, concerns remain around model governance, auditability and the use of opaque black box systems. Moody’s approach is to position AI as an orchestration layer rather than a replacement for established risk models. The validated analytical engines underpinning the analysis remain unchanged; the AI interprets user intent, identifies the relevant analytical components, co-ordinates outputs across different domains and translates the results into plain language.

“Instead of giving you just a probability of default [PD] number, we explain what that PD means to you, such as whether it’s higher or lower than previous years, or how it compares to that of your peers,” Bian says, by way of example. The platform prepares metrics to be “more decision ready” by connecting them to other relevant information that enriches the context to support a confident decision.

You don’t want your balance sheet interpreted by eight different teams in eight different ways. You want them to use the same foundational data, and to be able to draw multiple perspectives from that single source

Charlene Bian

The foundation is a domain-focused, multi-agent architecture sitting on top of Moody’s trusted data and engines, with specialised domain agents, such as credit analytics, balance sheet management, portfolio analytics and impairment modelling. A central orchestration layer co-ordinates their outputs into a single, auditable answer, traceable to the models and data behind it.

Relationship managers, finance teams, treasury functions and risk professionals can query Banking Decision Intelligence directly in conversation, as well as within a bank’s existing systems, rather than standalone as another tool.

It now supports a growing range of banking use cases, including loan committee preparation, risk reviews, relationship profitability analysis, concentration risk monitoring, liquidity and capital planning, CECL reporting, and asset-liability committee decision support.

Moody’s reports growing engagement across global, regional and specialist banks, with several pilot programmes under way.

“Imagine you are a CEO of a bank – you don’t want your balance sheet interpreted by eight different teams in eight different ways,” says Bian. “You want them to use the same foundational data, and to be able to draw multiple perspectives from that single source.”