DataGov-Factory: An Empirical Evaluation of Managed vs. Custom LLM Agent Architectures for Financial Data Governance

Key Points

• This research aims to evaluate the effectiveness of different LLM architectures in financial data governance.
• Controlled empirical comparison of three LLM deployment architectures
• Evaluation of platform-managed agents
• Assessment of custom multi-agent factory
• Analysis of direct API calls
• Platform-level prompt caching reduces context costs by approximately 90%
• Managed agents achieve 16% higher quality compared to other architectures

Abstract