How Balyasny Asset Management built an AI research engine for investing

OpenAI
Balyasny Asset Management built a centralized, AI-native research engine to accelerate complex investment analysis across its global teams.

Summary

Balyasny Asset Management (Balyasny), a global multi-strategy investment firm, established an Applied AI team to build an AI research system capable of reasoning, retrieving, and acting like a skilled analyst amid complex market data. The firm developed a rigorous evaluation pipeline, measuring models across over 12 dimensions, leading to the adoption of the GPT-5.4 model family as a reasoning engine alongside internal models. Key to their success was deep collaboration with OpenAI, designing for continuous feedback loops to rapidly improve model behavior, and implementing a centralized AI system with a "federated deployment" model, allowing local customization while maintaining universal compliance guardrails. This AI platform has significantly boosted efficiency, reducing multi-day research tasks to hours and enabling analysts to generate more structured, explainable insights, with approximately 95% of investment teams actively using the system.

(Source:OpenAI)