Raymond James uses knowledge graphs to ground financial AI agents
The financial firm deploys graph technology to ensure AI outputs match trusted data sources, improving accuracy for advisory tools.
Key points
- Raymond James Financial Inc. integrates knowledge graphs to support agentic AI infrastructure.
- The company uses Neo4j technology to connect disparate data domains for reasoning.
- Leaders say the approach ensures AI results are validated against golden sources.
Raymond James Financial Inc. is deploying knowledge graph technology to create a trusted infrastructure for artificial intelligence within its financial services operations. According to a report by SiliconANGLE, the diversified firm uses these graphs to connect data and ground AI agents in specific business expertise, helping to ensure accuracy in a regulated industry.
Jeff Griffith, vice president of AI engineering and data science at Raymond James, stated that using data graphs allows the company to harness AI power in dynamic ways. He noted that while users know how to retrieve routine information, graphs provide significant value when crossing domains or performing unique tasks that require connecting disparate data points.
Context and background
Raymond James utilizes Neo4j, a graph database platform, to represent relationships, dependencies, and business context that AI applications need to reason effectively. During a broadcast on theCUBE at the recent GraphSummit event, Griffith and Naveen Mahankali, vice president of technology at Raymond James, discussed how these technologies are changing the business landscape for the firm.
Mahankali explained that the firm has established golden sources and reliable systems that they use to validate AI outputs. By pulling data from different sources into the Neo4j knowledge graph, the team can run evaluations comparing the AI’s work against these trusted sources to ensure results are accurate before launching any new agent or tool.
Who is affected
The implementation of these AI and graph technologies is creating new efficiencies within Raymond James’ organization. Mahankali indicated that the firm sees future opportunities for AI to reduce the time and effort financial advisors spend sourcing information for clients. This shift aims to support advisors by providing accurate, grounded insights derived from the firm’s established data infrastructure.
What should you do?
Organizations exploring agentic AI should evaluate how to ground models in trusted data sources to maintain accuracy. Security teams should monitor how AI agents access and cross-reference internal data domains to prevent hallucination or data leakage. Companies should consider how graph technologies can help map relationships between data points for better reasoning.
What to do and how to stay safe: Raymond James
- Review current AI agent architectures to ensure they reference trusted, verified data sources.
- Implement validation steps that compare AI outputs against established golden sources before deployment.
- Map data relationships and dependencies to help AI systems reason across different business domains.
- Monitor how AI tools access cross-domain information to maintain accuracy and compliance standards.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
How does Raymond James ensure AI accuracy?
The firm uses knowledge graphs to pull data from various sources into Neo4j, where it runs evaluations against trusted golden sources to verify accuracy before launching agents.
What technology does Raymond James use for this?
Raymond James is a customer of Neo4j, using its graph technologies to represent relationships and business context for AI applications.
Who discussed this initiative publicly?
Jeff Griffith and Naveen Mahankali from Raymond James discussed the initiative with John Furrier on theCUBE during the GraphSummit event.
