AI in finance has a long way to go. Most applications fail to deliver on basic tasks, such as providing reliable stock prices, earnings data, and access to SEC filings. To change this, I built an open-source deep research AI agent for finance, inspired by Perplexity, but with Bloomberg-grade data access. This agent can answer complex research prompts end-to-end, providing structured research briefs, and even runs real code for on-the-fly analysis.
What sets this apart is its ability to access a vast amount of data, including SEC filings, earnings, balance sheets, market data, insider transactions, financial news, and even peer-reviewed finance journals. The agent can produce plots and charts, and returns sources and tables for verification.
I built this using a single search API that covers all the finance data and news, allowing the agent to query in natural language. The tech stack includes Next.js, OpenAI GPT-5, Valyu DeepSearch API, and Daytona for code execution. I’d love for the community to try it out and provide feedback on the workflow of the agent. The goal is to create a powerful tool that can truly make a difference in the world of finance.