From data to decisions: how LSEG is scaling trusted AI

TL;DR: LSEG partners with OpenAI to scale trusted AI by grounding models in high-quality financial data, shrinking release cycles from months to two weeks, and expanding the analyst role from data processing to deeper insight. ## Trust Starts with Data LSEG sits at the heart of financial markets, providing trusted information and services across the ecosystem. The company’s approach to AI begins with a fundamental principle: trust in AI starts with trust in data. > “Any product that we’re building, we’re assessing not only from the data accuracy standpoint, though, we’re also assessing the groundedness of the response. We’re assessing the way the model is reasoning through the financial information. We’re assessing the data surplus. We’re assessing the data fidelity. We want to make sure all the data and analytics that we’re building is as consumable by humans and by AI as well.” A lot of LSEG’s customers had already been doing heavy lifting on how data models work—how to align, rebase, and consume them in a scalable way. LSEG’s role is now to offer that trusted data in a form that AI applications can use directly. ## The Role of MCP and ChatGPT Integration Many of LSEG’s clients are also OpenAI users, running ChatGPT and other tools. Through LSEG’s Model Context Protocol (MCP), the company is able to provide its trusted data and services directly inside ChatGPT. > “We have many clients who are using OpenAI as well. And they use things like ChatGPT and through capabilities like LSEG’s MCP or model context protocol, LSEG is able to provide its trusted data and services to those customers through ChatGPT.” This integration means analysts can query ChatGPT and get grounded, high-fidelity financial information without leaving their familiar interface—all backed by LSEG’s governance and quality standards. ## Accelerating the Operating Model with OpenAI OpenAI is a crucial partner for LSEG. The two teams began working closely to see how OpenAI could help accelerate LSEG’s operating model and move forward faster together. > “OpenAI is a crucial partner. We started working very, very closely with OpenAI team to see how OpenAI can help us accelerate our operating model and how we can move together faster.” That collaboration has made it possible to scale best practices more easily while still embedding the standards and skills required to maintain a high bar for trust and safety. ## Shrinking Release Cycles for AI Products One of the most tangible changes has been in the speed of delivery. Historically, release cycles ranged from three to six months, depending on the product and what was being built. With the shift to AI, LSEG shrunk the cycle to two weeks. > “Now, to all of our AI products we shrink the release cycle to two weeks. So now we’re releasing the products on a bi-weekly basis. The information moves faster and the decision process is a lot faster and therefore the iterations are all faster.” This bi-weekly cadence allows LSEG to respond to market changes and customer needs far more rapidly than the old model. ## Empowering Analysts and Scaling Best Practices Speed isn’t just about technical releases—it changes the nature of the work itself. The role of the analyst can now expand substantially. > “The role of the analyst now can expand substantially into other things. There is much more opportunity for them to do much deeper research, to do much more orthogonal insights in a way that they may not have had the time to do before.” By automating routine data processing and grounding with trusted AI, analysts can focus on novel connections and strategic thinking. LSEG is also thinking about how to scale these best practices across the entire organization. > “The collective power of 27,000 employees really leaning into the power of AI. It is extraordinary to imagine what that looks like. That is really exciting where we can move fast in a safe scalable way and take advantage of some of the frontier capabilities of OpenAI.” ## Source Watch the full video: [From data to decisions: how LSEG is scaling trusted AI](https://www.youtube.com/watch?v=sU9-u5p-jA0)