OpenAI's chief financial officer Sarah Friar has set out five lessons from building what she calls an "AI-native" finance function, in a piece published on the company's site.
Friar said OpenAI is working towards a "zero-day close" and continuously updated, automated forecasting, aiming to give leaders a real-time, traceable view of the company's financial position rather than periodic snapshots.
Among the lessons: give staff broad access to AI tools before mandating use, then let hackathon-style experimentation surface real applications. One example cited is IR-GPT, a custom tool grounded in approved investor relations materials that can produce a first draft response to diligence questions in seconds.
Friar cited internal OpenAI research showing 40% of finance professionals' specialised AI use falls outside traditional finance tasks, with 22% involving engineering work. She said finance staff are increasingly building their own dashboards and tools using ChatGPT Work and Codex, including one non-coder who built a tool converting monthly advertising forecasts into daily and weekly plans.
Friar also stressed accountability, arguing CFOs should set clear rules on data access, approval thresholds and escalation, and should measure AI's value by whether work was completed reliably and cost-effectively, not by seats or tokens used.
