It’s time to turn the hype about AI at your company into real outcomes. While many companies spent the past year talking about AI, a team of innovators at Moody’s, a leading risk assessment firm, developed and quickly launched their custom enterprise copilot to thousands of employees. The approach drove productivity and innovation across the company and led to a series of AI product launches throughout the year.
Let’s walk-through how Moody’s strategically dove into AI so you can quickly build your custom copilot today on Teams leveraging best in class pre-built AI components from Azure OpenAI and Teams AI Library boosting productivity and immersing your organization in AI.
Generative AI in Financial Services with Moody’s copilot
How to begin my enterprise’s AI transformation
Start by empowering your employees
Two of the top questions that every Microsoft customer asks Satya Nadella, Microsoft’s CEO, is not only how, but how fast, they can apply the newest generation AI to address the biggest opportunities and challenges they face. As generative AI has numerous applications with tremendous potential, it can be hard to know where to start.
Moody’s CEO, Rob Fauber, provides a roadmap on how to successfully transform your organization with AI quickly by integrating it directly into the workflow of 14,000 employees on Microsoft Teams.
“We recognized early on that our people would be the driving force of our innovation.
So rather than a traditional, top-down ‘corporate approach,’ we deployed the technology directly to our 14,000 global employees — or 14,000 innovators, as I call them — and asked them to tell us how best this technology could create value.
And — wow — was that the right call — over the past few months we’ve seen hundreds of ideas for different use cases, workflows and GenAI-enhanced products.
This technology is a game-changer.” – Rob Fauber, president and CEO of Moody’s Corporation
Fauber’s bottom-up strategy to AI capitalizes on every company’s most important source of innovation, its people. Starting your AI transformation with a custom copilot enables every employee to become an innovator as they learn to optimize their workflows with AI over Teams.
When employees use a custom copilot to assist with crafting a sales pitch, catching up on a missed meeting on Teams, or analyzing vast amounts of financial information, they are improving their productivity while gaining insight into how AI can solve other problems across the organization. This approach gave Moody’s an edge when developing premium customer-facing AI products as they were able to innovate and lead customers from experience.
How can I accelerate my AI development
Leverage Azure OpenAI and Teams AI Library
To position Moody’s as an AI leader in its industry, it had to act swiftly in deploying their custom copilot.
Facing the task of quickly innovating to seize a once-in-a-generation opportunity, Trevor O’Brien, Moody’s Sr. Director of Technology Innovation, and his team successfully launched Moody’s copilot in less than 30 days despite starting from a blank canvas.
This accomplishment was made possible through deliberate development stages and effective utilization of Microsoft AI services and tools like Azure OpenAI and Teams AI Library:
- Phase 1: Leverage pre-built AI components from Azure OpenAI and Teams AI Library to accelerate your custom copilot’s development and launch it globally via Teams.
- Phase 2: Ground your custom copilot’s answers in key internal and external data sources for tailored insights.
- Phase 3: Add sophisticated features, more data sources, and complete the roll-out to the entire organization.
Phase 1 – Expedite development by leveraging pre-built AI components from Azure OpenAI and Teams AI Library
“We went from nothing. We had a blank canvas with no code, no Teams app, no data to our first product in the hands of 4000 users in less than 30 days. […] I would say a lot of that is because we were able to leverage the pre-built components like Teams AI Library” – Trevor O’Brien, Moody’s Sr. Director of Technology Innovation
This phase was crucial for laying the groundwork and discovering the best use cases for the company to further iterate and customize Moody’s copilot beyond the typical Chat GPT like experience. The goal was to quickly deliver a secure and effective tool that could provide fundamental services such as specialized financial analysis, processing documents, and summarizing content.
To accelerate development, Moody’s leveraged market-proven AI building blocks to power Moody’s copilot. Instead of spending years building their own large language model (LLM), Moody’s used Azure OpenAI Service to select a pre-built GPT large language model to power their AI experience. For seamless and fast integration into employees’ workflow on Teams, Moody’s developers leveraged Teams AI Library for pre-built component scaffolding like adaptive cards for an effective user interface enabling the copilot to all users across their chats, channels, and meetings.
Leveraging pre-built AI services and components in every step of the development process resulted in a high-quality initial product instilling the confidence for a significantly wider launch than normal to over 4,000 employees in less than 30 days.
Phase 2: Improve productivity and insights by grounding your custom copilot in your data
“A lot of what Moody’s’ employees are doing is searching for data across different data sets or documents. And what we’re seeing is improved productivity and being able to get to the interesting nuggets of information about a company, about a given document that you’re searching in, and getting the information that you’re looking for quickly with this product.
[…] A big part of the iteration, was allowing users to anchor their answers in data, not just giving the vanilla GPT functionality. Part of the early feedback was a desire for more features there.” – Louise Hopkins, Moody’s Director of Product Management
Moody’s employees conduct intense research, digging through a vast network of internal and external datasets including information on 445+ million private businesses worldwide to evaluate the risk factors and provide valuable insight to their customers. The process of searching through multiple documents and numerous data sources to connect the pieces in a contextually relevant narrative is time intensive and cognitively demanding. This pain point is not exclusive to Moody’s as a McKinsey report, discovered knowledge workers spend around 1.8 hours every day – 19% of their time searching for and gathering information.
Grounding an LLM in internal and external datasets and enabling uploading documents connects disparate information sources allowing users to “chat with their own data” streamlining the search and analysis process. Instead of spending hours digging through a dozen documents and databases, Moody’s employees improved productivity by using Moody’s copilot to find and intelligently analyze the most relevant information instantly solving a longstanding corporate inefficiency of finding the right information in an ocean of data.
Best of all, grounding your LLM with your data is straightforward with Azure OpenAI and the Teams AI Library. Without re-training or fine-tuning the model, simply connect the data sources to your LLM using Azure OpenAI on your data or Retrieval Augmented Generation with the Teams AI Library. The data remains stored in the data source and location you designate. No data is copied into the Azure OpenAI service. When a user prompt is received, the service retrieves relevant data from the connected data source and augments the prompt. All of this can be set up with a few clicks.
If your organization has massive amounts of data, similar to Moody’s, connecting different data sources can be done in phases targeting the highest quality and most pertinent sources to your use case first.
Phase 3 – Deeper customization and multi-faceted analysis
With heavily regulated and less regulated sides of the business, Moody’s used the Microsoft Teams ecosystem and Azure Active Directory to help Moody’s copilot adjust the information it serves to users based on where the employee works inside of Moody’s. The custom copilot was connected to detailed policy information enabling it to answer questions from employees along with linking the source of that information and related policies for added transparency. These types of policy and data security related features allowed for the full product roll-out to Moody’s 14,000 employees including their most regulated departments.
As further data sources were linked to the custom copilot, its value increased by enabling real-time multi-faceted analysis, which has not been possible at this scale, quality, and speed until now.
“One of the things that Moody’s copilot enabled us to do is to reach into many different products including product data sets that haven’t been connected in a singular interface before. It creates this new experience where our 14,000 internal innovators can ask a question and get a response from multiple Moody’s products in an adaptive card. We’re using the Teams AI Library to go and reach static information as well as live feeds like news and other kinds of real time content, and then send all of that to the LLM as context to get back a summarized response. It’s actually allowing us to create a new type of product that is a combination of a lot of our different product sets.” – Trevor O’Brien, Moody’s Sr. Director of Technology Innovation
The beauty of a custom copilot with multi-faceted analysis is that everything required to build it is readily available. The models in Azure OpenAI, the Teams user experience through the Teams AI Library, and all your corporate data sources are waiting to be connected and provide unique insights to employees.
Beyond enhancing insights and productivity, crafting a custom copilot can ignite AI-driven innovation across your company. Moody’s recently built off its internal copilot with a premium customer-facing offering, Moody’s Research Assistant. This tool offers Moody’s clients a swift means to aggregate and distill complex data from Moody’s extensive repository of research, data, and analytics, mirroring the internal capabilities provided to Moody’s own employees by its custom copilot. Research and analysis that used to take days can now be accomplished in minutes, freeing up time for strategic decision-making and faster execution.
These products came into being not from talking about AI, but by diving into it as an organization and utilizing the best services and tools available to build AI. You can start by building your custom copilot the same way today.
Find out more about Moody’s on LinkedIn.
Additional resources
Ready to harness the power of the Teams AI library and create your app’s custom copilot for Teams?
Get started here:
0 comments