Franklin Templeton uses OpenAI to support investment research
Franklin Templeton is leveraging OpenAI tools, including Codex, to transform its investment research process by building a 'connected intelligence layer' that integrates continuous AI insights with human judgment. This approach moves the firm beyond traditional, sequential human-led analysis by surfacing actionable insights from diverse data sources for portfolio managers.
Creating Connected Intelligence Layers
Franklin Templeton's implementation of AI began by connecting models to a database, but initial results were insufficient because the models did not understand the relationship between the data points. The solution involved introducing agents and a semantic layer—a structured model that defines what the firm's data means and how all components are interconnected. This 'connected intelligence layer' allows the firm to bring complex data into the tools and surface meaningful insights.
This integrated approach aims to unify various perspectives on an investment, including research, risk assessment, compliance, and trading. The ultimate goal is to consolidate these disparate inputs and present synthesized insights directly to the portfolio managers who make investment decisions.
Workflow Transformation in Investment Research
Historically, the equity-research workflow was sequential and human-led: an analyst would perform initial stock analysis, review earnings calls, and study competitors within each business line. Franklin Templeton is moving away from this model toward a vision where its investment teams have access to a governed, reusable intelligence foundation. This foundation combines trusted connectors, portfolio and workflow context, and agentic capabilities to surface actionable insights for both analysts and portfolio managers.
This transition represents a fundamental shift from simply augmenting the existing research process to actively reimagining the entire investment research workflow.
Principles for AI in Asset Management
Franklin Templeton emphasizes that standard AI applications, which often reinforce existing user preferences, are unsuitable for asset management. The firm requires its AI systems to prioritize diversity of opinion and comprehensive risk analysis.
- Ensuring Diversity: The AI must counteract the tendency to narrow perspectives by ensuring that investment teams access a broad diversity of research and opinions, preventing potential bias from a single research provider.
- Risk Focus: The AI is designed to emphasize investment risks. A well-reasoned investment idea must be paired with an analysis of its associated risks, making portfolio managers aware of potential downsides. Identifying investment risks within a research note is considered a positive outcome.
Best Practices for AI Adoption
Franklin Templeton's experience highlights several best practices for introducing and scaling AI within a regulated financial environment:
- Encourage Experimentation within Guardrails: Leadership encourages innovation and experimentation with AI, provided it occurs within clearly defined risk guardrails. Each initiative should be viewed as an opportunity to strengthen capabilities and inform future decisions.
- Rapid Tool Deployment: The firm quickly brings practical labs to teams following new technology launches to allow for immediate testing and integration into daily work.
- Focus on Understanding: The goal of implementation is to ensure that teams fully understand the power and capabilities of the technology they are using.
- Build with Appropriate Controls: The path from experimentation to production-grade tools requires robust controls, including security reviews, expert support, and strict data agreements.
Ultimately, the long-term goal is to make this connected intelligence accessible across all investment teams, ensuring that the essential human perspective—such as analysts speaking with various experts and executives—remains central to the decision-making process.