Balyasny Asset Management, a massive multi-strategy investment firm, has created its own AI research engine to help analysts sift through mountains of financial data. The firm set up a dedicated Applied AI team of 20 experts in late 2022 to build tools that work like a skilled analyst — reasoning, retrieving information, and taking action. Before going live, they put models through grueling tests across 12 dimensions, including forecasting accuracy and handling messy data. GPT-5.4 came out on top for planning and multi-step tasks, so it now powers the system alongside internal models. The result? Deep research that used to take days now wraps up in hours, and 95% of the investment team uses it. They even brought OpenAI in to observe real user workflows, fine-tuning the experience. Why it matters: This shows how custom-built AI, not off-the-shelf tools, can give financial firms a serious edge in speed and precision — and it’s a blueprint for other industries drowning in data.