Franklin Templeton is building an AI-supported equity research system that would organize data through a semantic layer, helping portfolio managers examine investment ideas and related risks for review.
The firm is applying the approach to equities first. Researchers have traditionally handled work in distinct steps, such as assessing stocks, reviewing earnings calls and comparing competitors. The planned setup would connect trusted data with portfolio records and workflow details, giving research teams a reusable foundation.
Katrina Dudley, Franklin Templeton’s senior vice president of Public Market Investments, said an early database trial left AI models unable to make sense of links among records. The team responded by adding a semantic layer to map those relationships.
The developing system would bring research, risk, compliance and trading perspectives into a shared workflow. It is intended to show portfolio managers different views of investment ideas and surface related risks for review. The software would arrange information for portfolio managers to consider; they would remain central to investment decisions.
Hackathons have brought investment staff and technology specialists together to work with OpenAI’s Codex. Investment AI champions at the firm are advancing prototypes toward production use while addressing security reviews and data controls. Franklin Templeton has not identified which Codex version it used.
The firm has not provided performance figures or a staffing count for deployment.
