Nov 18, 2026 | AI and product innovation
Trusted data, faster decisions: how Thomson Reuters turned its finance data into decision intelligence using Microsoft Fabric
Thomson Reuters Monthly Business Review used to start the same way most do. Leaders in the room, a deck on the screen, and the first ten minutes spent aligning on whether the numbers were right. Today it starts differently. Everyone opens the same app, every figure traces back to its source, and the meeting begins with a single question: what do we do about it?
That shift did not happen because the data got simpler. Thomson Reuters has more data than ever, more tools to analyse it, and more people who need to use it. What changed is that the data became trustworthy enough to act on in real time, without a team of analysts standing between a question and an answer.
Trusted data is the foundation. Thomson Reuters DeepVista is an AI framework built to put trusted data in the hands of employees when they need it. Every answer comes from certified semantic models on Microsoft Fabric, so employees get consistent, trusted answers, whether they’re using a dashboard, an app, or an AI agent. The same semantic models and business context power Thomson Reuters DeepVista’s AI agents in Teams and the Monthly Business Review app, built with Fabric apps. This means leaders get consistent answers whether they’re exploring the app or asking questions in a conversation with the AI agent.
The Monthly Business Review is powered by Fabric apps. Built with Rayfin and deployed on Microsoft Fabric, the app gives Thomson Reuters a faster way to turn trusted data into an interactive experience for its Monthly Business Review process. Every value traces back to its source and latest refresh, and teams can add their own notes and context directly in the app. No digging through reports. No waiting on another team.
Adoption shows trust. Since launching in June 2026, Thomson Reuters DeepVista has grown from zero to thousands of active users who have asked thousands of questions. People kept coming back because they trusted the answers.
The problem with data at scale
Thomson Reuters makes products for professionals who need to get things right. These include lawyers, tax professionals, and compliance teams. We hold ourselves to the same standard internally, which meant that when we looked at how our own people were working with data, we asked the same question we ask our customers: is this actually working?
Thomson Reuters has no shortage of data. What we needed was a way to make it consistent and immediate across the whole organisation. In a business of our size, the same metric can mean something subtly different depending on which system or team you ask. It’s what happens when organisations grow. But the consequence is that data slows decisions down rather than accelerating them.
Two things came together to make this the right moment to act. Advances in governed semantic models and AI made it technically possible to put trusted data to work in real time. And we had a clear organisational need: give people a reliable, consistent way to understand business metrics without navigating competing systems or waiting on analyst support.
What We Built
The platform runs on two layers. The Finance and Enterprise Semantic and Contextual Layer creates one consistent version of our most important business data. So, when someone asks about revenue, headcount, customers, or product performance, everyone is working from the same answer. Building that consistency across an organisation of our scale was not simple. On top of it sits the Fabric IQ Layer, which encodes business context around those numbers, not just what the data says, but what it means for this company, right now. That’s what turns a figure into something you can act on.
We went from pilot to enterprise deployment within a single fiscal year, a timeline that reflects both the urgency of the need and how quickly the platform proved its value.
Where people actually work
Both layers feed into AI agents and Copilot capabilities that live inside Microsoft Teams and Microsoft 365, where our people already spend their working day. There’s no new system to log into, no interface to learn. An employee asks a question in natural language, through the tools already open on their screen, and gets an answer.
“This is a breakthrough in how Finance works. Certified semantic models give us numbers we can trust. Thomson Reuters DeepVista’s AI agents let analysts and leaders ask questions in natural language and get insights they can act on. The monthly business hub solution automates the business review, so we spend our time on what to do next,” says Michael Goddard, Head of Corporate Finance, Thomson Reuters.
Why we built on Microsoft Fabric
Microsoft Fabric is a next-generation data platform built to accelerate AI innovation, making it easier for people across the company to use and interact with data. With Fabric IQ, we can build certified semantic models once and reuse them across dashboards, self-serve tools, AI agents, and MCP Servers. Its integration with Microsoft Entra ID and Microsoft Teams carries each user’s identity and row-level security all the way through the stack, so we could scale from pilot to enterprise without duplicating data or compromising governance.
“With Thomson Reuters DeepVista, Thomson Reuters has set a benchmark for how large organizations should approach enterprise data and AI. They built trusted semantic models once and reused them as business context across every experience, powering Microsoft Copilot, AI agents in Microsoft Teams, and even custom applications built with Fabric apps. Thousands of employees now get consistent answers, with governance and security following the data wherever it goes. Grounded in Fabric IQ, DeepVista gives their teams one shared understanding of how the business operates. This is the kind of enterprise transformation we designed Fabric to make possible,” says Bogdan Crivat, Corporate Vice President, Microsoft Fabric.
What comes next
We are now extending Thomson Reuters DeepVista’s foundation across other parts of the business. But the ambition goes further than that.
Helping people ask better questions of their data is a starting point, not the destination. The next step is AI agents that understand business context well enough to surface what matters before anyone must go looking, shifting the organisation from reactive to proactive, from data-rich to genuinely decision-ready.
Our ambition is to make trusted intelligence available at the point of every decision, creating an enterprise where data doesn’t just inform people, but continuously helps them make faster, smarter decisions.