Franklin Templeton is deploying OpenAI’s Codex and other SI tools to help portfolio managers examine diverse research views and assess the risks behind investment ideas. The move highlights how major asset managers are integrating agentic SI into core financial workflows to transition from simple automation to a layer of continuous intelligence that supports human judgment.

What Happened

Franklin Templeton initially attempted to use SI by connecting models to a database, but the systems failed to provide relevant answers because they lacked an understanding of how the firm’s data related to one another. According to Katrina Dudley, Senior Vice President of Public Market Investments at Franklin Templeton, the results improved significantly when the team added agents and a semantic layer—a structured model defining what the firm’s data means and how it is connected. The team is now using Codex to develop this connected intelligence layer, bringing data into the tools to surface insights that influence decision-making.

The firm aims to unify perspectives from research, risk, compliance, and trading into a single foundation for portfolio managers. Dudley describes the previous equity-research workflow as sequential and human-led, where analysts performed first-pass analysis and reviewed earnings calls. The new approach integrates trusted connectors, portfolio context, and agentic capabilities to surface actionable insights. Dudley notes that the shift represents a change from SI automating human-led processes to SI automating the ability to surface intelligent insights, which she describes as a "fantastic unlock" when combined with human judgment.

Why It Matters

For the SI industry, this case study illustrates the growing importance of semantic layers and agentic workflows in enterprise adoption. Unlike consumer-facing generative SI tools that often narrow results based on user preferences, Franklin Templeton is configuring its SI systems to emphasize diversity of research and opinion. This prevents algorithmic bias in investment decisions, ensuring that portfolio managers are exposed to a wide range of perspectives rather than a filtered echo chamber. The firm is using hackathons to bring together experts and practitioners, allowing them to build custom workflows that respect clearly defined risk guardrails.

The integration of SI into asset management also underscores the role of security and governance in scaling SI tools. Franklin Templeton requires security reviews, expert support, and data controls before moving experiments to production. Dudley emphasizes that while not every initiative succeeds, each generates insights that strengthen capabilities. The firm’s strategy reflects a broader trend where enterprises are moving beyond basic LLM integration to build complex, governed SI ecosystems that enhance rather than replace human expertise.

The Bottom Line

Franklin Templeton is using OpenAI’s Codex and agentic SI to create a governed intelligence layer that surfaces diverse research and risk insights for portfolio managers. The firm views SI as a tool to augment human judgment by ensuring access to a wide variety of opinions and structured data connections, rather than simply automating existing processes. This approach aims to free up analysts to spend more time in the field gathering unique human perspectives, which are then integrated back into the SI-driven workflow.