How Schroders built its multi-agent financial analysis research assistant

Schroders, a leading global active investment manager, wants to enable its analysts to focus on strategic thinking by automating data collection and research. To achieve this, Schroders collaborated with Google Cloud to build a multi-agent research assistant prototype using Vertex AI Agent Builder. The prototype aims to reduce the time spent on company research from days to minutes, allowing analysts to screen more companies and discover promising investment opportunities. A multi-agent system was chosen due to its ability to specialize in specific tasks, scale, and orchestrate complex workflows. The system consists of various agents, each designed to perform a specific task, such as R&D or working capital analysis. Vertex AI Agent Builder was selected as the core platform for developing and deploying the multi-agent system, providing benefits such as access to state-of-the-art Google foundation models and pre-built connectors for various tools and data sources. The system allows for intuitive, flexible end-user interaction, with analysts able to create and configure agents, combine them into workflows, and personalize their interactions. The system's architecture is designed to ensure each component focuses on its strength, providing vital flexibility and encouraging user adoption. To measure success, Schroders implemented a multi-faceted evaluation strategy, including automated metrics, human-in-the-loop review, and a ground truth dataset. The collaboration demonstrated the feasibility of developing an equity research AI research assistant that can significantly enhance analyst productivity.