Jefferies, a global investment bank, introduced an AI-powered trade assistant to enhance real-time data analysis for its equities trading desks. The solution enables traders to access insights from vast datasets without coding or waiting for IT support. The system leverages AWS technologies to provide accurate, timely insights directly within the trading environment.
The AI trade assistant uses Amazon Bedrock, large language models, and Model Context Protocol (MCP) tools to securely connect to data sources like trade repositories and FIX messages. When traders submit queries, the system interprets natural language intent, generates SQL queries, and executes them against relevant data sources. The conversational interface allows traders to explore insights dynamically, with the system maintaining context across the session to provide relevant suggestions.
The solution integrates with Jefferies’ existing trading infrastructure through an eight-step process, including authentication via Amazon EKS and secure data access through row-level filters. Amazon Bedrock Guardrails ensure compliance with policies by filtering personally identifiable information and preventing unauthorized data access. The system also logs conversations for audit trails, meeting regulatory requirements.
Source: awsml
