Franklin Templeton has announced the development of AI-driven workflows to enhance its investment research processes, starting with equities, leveraging tools from OpenAI including its Codex platform to help portfolio managers compare diverse research views and analyze investment risks. According to Katrina Dudley, senior vice president of Public Market Investments, the initial approach of simply connecting AI models to a database did not yield satisfactory results due to the models' inability to understand relationships within the data, so the team introduced a semantic layer that structures data to reflect its meaning and interconnections. The enhanced system integrates trusted connectors, portfolio context, and workflow context to build a reusable intelligence foundation designed to combine research, risk, compliance, and trading perspectives while exposing portfolio managers to diverse opinions and explicitly highlighting risks, with human judgment remaining central to decision-making. Codex was originally launched on May 16, 2025, as a cloud-based software-engineering agent powered by the codex-1 model, an OpenAI o3 variant trained through reinforcement learning on real-world coding tasks; while the specific version used by Franklin Templeton has not been disclosed, the firm has employed the tool in hackathons pairing technical experts with investment practitioners, supported by its Investment AI champions who work to transition prototypes into production-ready workflows through rigorous security reviews and data controls.