EU Macroeconomic Analysis with Deep Agents and You.com

Macroeconomic research teams have long spent considerable time deriving clearly sourced briefings from raw data due to the heterogeneity and fragmentation of country-level GDP data. To address this, an agent architecture was built combining Deep Agents, LangSmith, and the You.com Finance Research API.

This system analyzes 2025 GDP data for the EU's 27 member states to detect anomalies and investigate structural and cyclical factors by sector. The entire execution takes approximately 45 minutes, with API call costs totaling about $2.20. The You.com Finance Research API combines licensed structured data from sources such as S&P Global with real-time web intelligence, achieving a score of 87.29% on FinSearchComp, a financial services benchmark.

The analysis revealed that Ireland exhibited the largest anomaly, with a GDP growth rate of 12.3% driven by a surge in pharmaceutical exports in anticipation of US tariffs. The industrial sector contributed +6.55pp, but Modified GNI showed a much lower figure. In contrast, Germany was classified as the opposite anomaly due to structural contraction caused by reliance on the automotive industry and a construction downturn.

In the financial services industry, the process of deriving conclusions is considered as important as the conclusions themselves. LangSmith records all queries, responses, and intermediate results generated during agent execution to preserve the decision log. This ensures that every finding in the final report is traceable back to the primary source that generated it.