How Jefferies Optimized Front-Office Trading Operations with an AI Assistant on AWS

Investment bank Jefferies has developed an AI assistant for front-office traders based on AWS. The solution uses the open-source SDK Strands Agents, Amazon Bedrock (with the Anthropic Claude model), and Amazon Bedrock Knowledge Bases with Amazon Titan embeddings. The assistant is integrated via MCP tools with various data sources: in-memory databases, SQL stores, and FIX messages. Traders can ask questions in natural language, and the LLM-based agent generates SQL, executes queries, and visualizes results. The system ensures security through Amazon Bedrock Guardrails, PII filtering, and row-level access control. Since launch, the solution has improved efficiency, reduced IT workload, and democratized data access. Jefferies plans to deploy the assistant globally.