Franklin Templeton turns to OpenAI for connected investment research

Building a connected intelligence layer

Franklin Templeton initially connected AI models to data stored in a database, but the approach did not produce the answers its teams needed. According to the firm, the models lacked an understanding of how the data fit together.

The investment company then added agents and a semantic layer, which it describes as a structured model of what its data means and how it is connected. Franklin Templeton is using Codex and other OpenAI tools to develop this connected intelligence layer and bring data into workflows that can surface useful insights.

Research, risk, compliance and trading can each provide a different perspective on an investment. Franklin Templeton wants to bring those inputs together for portfolio managers, who make the investment decisions. Katrina Dudley, senior vice president of public market investments at Franklin Templeton, said the firm has been able to build what it calls a “connected intelligence layer.”

From sequential research to role-specific workflows

Dudley describes the earlier equity-research process as human-led and sequential. An analyst would perform a first-pass stock analysis, review earnings calls and study competitors in each business line before presenting a new idea to portfolio managers or updating them on an existing position.

Franklin Templeton’s intended workflow would give investment teams access to a governed and reusable intelligence foundation. The foundation is designed to combine trusted connectors, portfolio and workflow context, and agentic capabilities that surface actionable insights for analysts and portfolio managers.

The firm has used Codex in hackathons, where experts work alongside investment practitioners to develop tools for their own roles. Its Investment AI champions are helping move experiments toward production-ready workflows, with security review, expert support and data controls included in the process.

“We have moved from augmenting our research process to reimagining the investment research workflow,” Dudley said.

AI designed to widen perspectives and highlight risk

Franklin Templeton says some AI behaviors that may be useful elsewhere could create problems in asset management. Systems that learn a user’s preferences and show more of the same could narrow the research perspective, such as repeatedly surfacing reports from a favored provider.

The firm instead wants its systems to provide diversity of research and opinion. It also wants AI to emphasize the risks behind an investment idea, rather than presenting an idea without the analysis needed to assess it. Dudley said an investment idea should be well reasoned, while also making portfolio managers aware of its risks.

For users, the approach is intended to support analysts and portfolio managers without removing human judgment. Franklin Templeton says it is encouraging experimentation within clearly defined risk guardrails, while using practical labs, training, security reviews and data agreements as teams develop new tools. Dudley’s longer-term goal is to make the connected intelligence available across investment teams, while analysts continue speaking with experts and professionals and bringing that human perspective back to their teams.