Building an AI-Native Finance Team

OpenAI scaled its finance team from the ground up, making AI central to its workflows, decision-making, and business support. The goal was to build zero-day close for real-time book closing and continuous forecasting that automatically updates.

Zero-day close provides a real-time, adjustable, and traceable financial state, while continuous forecasting reveals business changes, future possibilities, and decisions that can alter outcomes. To achieve this, the team is transitioning from static spreadsheets and manual search/reporting tasks to real-time tools.

The first lesson is to provide all team members with AI access while also creating opportunities to experiment and solve real problems. The finance team held hackathons with sales engineers to identify repetitive tasks and developed a custom GPT, IR-GPT, which answers due diligence questions based on investor relations approval materials. They have also begun building custom GPTs for procurement and tax areas.

The second lesson is that entire workflows leading to decision-making must be redesigned, not just individual tasks. To reduce repetitive work in forecast reviews or closing processes—such as finding data, adjusting spreadsheets, explaining variances, and creating charts and documents—the team is pursuing a method to connect the following data into a single flow:

• Approved spending plans
• General ledger actuals
• Purchase orders
• Accrued expenses
• Transaction details

OpenAI emphasizes that the impact of AI adoption lies not in simple technology adoption. Work must be redesigned around critical decisions, autonomy for experimentation must be guaranteed, accountability must be built into each workflow, and tasks reliably completed by AI must be measured to secure clear ROI.