Jefferies developed a trade assistant to optimize its front office trading operations, according to a post on the AWS Machine Learning Blog. The solution is built on Strands Agents, an agent harness SDK for creating AI agents that can reason, plan, and act by orchestrating calls to foundation models and external tools.
The system uses large language models, Amazon Bedrock, and Amazon Bedrock Knowledge Bases. It also incorporates the Model Context Protocol (MCP), an open standard that lets AI agents connect securely to varied data sources and tools through a unified interface. The post outlines the solution architecture, the reasoning behind the chosen technology stack, lessons learned during implementation, and the business impact realized at Jefferies.
Why it matters
The case describes how an agent-based approach combining foundation models with external tool orchestration can be applied to trading operations, illustrating the use of open standards like MCP for secure data connectivity in a financial context.
Who should care
Enterprises in financial services and teams evaluating agent frameworks, Amazon Bedrock, or MCP for building tool-connected AI agents may find the described approach relevant.