128 From AI experimentation to reliable operation INTERVIEW LSEG’s Nej D’Jelal explains to Lindsay James how financial services firms are moving beyond AI experimentation towards trusted, auditable and increasingly agentic workflows The financial services industry has spent two years exploring generative AI in regulated workflows. For Nej D’Jelal, group head of workflows, data and analytics at LSEG, that journey is now entering a new phase. “At first, firms focused on experimentation: identifying use cases, establishing governance and building the technology stack required to make AI useful,” he says. “That meant ensuring consistent access to data and systems capable of answering questions in ways that made sense to end users. There was also an element of fear of missing out as organisations sought to understand industry developments.” But over the past 12 months, and particularly the last six, D’Jelal has seen real conviction and transition from that experimentation into adoption. “Workspace AI Search now has 17,000 active users and Deep Research has reached 7,000 users, with adoption quadrupling since Q1,” he says. For financial professionals, that trust is critical. “They make trading and investment decisions,” D’Jelal says. “There’s a real downside to making the wrong decision.” Generative and agentic systems are powerful because they can interpret a question asked in plain English, understand its intent and map it to a broad range of data and capabilities. Yet their probabilistic nature means they cannot provide certainty alone. LSEG addresses that challenge by combining AI’s ability to synthesise information with deterministic tools and trusted and licensed data, including proprietary market data, research, Reuters News, transcripts, filings and events. “The certainty and the proprietary and exclusive nature of the underlying data is so critical to our users,” D’Jelal says. Transparency is equally important. When LSEG delivers an insight through its user interface, it aims to let customers source the transcript words supporting it. This enables them to validate, share and act on the insight. “Auditability is a critical design principle that we have factored into the way in which we build these products,” D’Jelal says. That principle is central to LSEG’s Workspace AI Search and Workspace Deep Research. Workspace AI Search supports everyday financial intelligence across market data, filings, Reuters News and research, while Deep Research tackles more complex questions by coordinating structured and unstructured content, connecting evidence and producing transparent, traceable research outputs. Behind both solutions are agents that orchestrate questions across LSEG’s data, from macroeconomic information and company estimates to transcripts and filings. Beyond Workspace, those capabilities are increasingly available within Microsoft Teams and other customer workflows, while maintaining
RkJQdWJsaXNoZXIy NzQ1NTk=