OpenAI outlines an AI-native model for finance operations
OpenAI is building an AI-native finance function around two long-term ambitions: a zero-day close and automated, continuously updated forecasting. Sarah Friar, OpenAI’s CFO, said the company is still progressing toward both goals after building a finance team to support rapid growth.
The zero-day close is intended to give leaders a real-time, reconciled and traceable view of the company’s financial position. Continuous forecasting would build on that base by showing how the business is changing, what may happen next and which decisions could alter the outcome.
From periodic reporting to live decision support
OpenAI describes the traditional finance workload as a chain of manual tasks: finding current data, reconciling spreadsheets, explaining variances, preparing documents and turning them into presentations. Its proposed operating model redesigns the workflow around the decision rather than the preparation of inputs.
For the close, OpenAI is working to connect approved spending plans, general-ledger actuals, purchase orders, accruals and transaction details in a continuously reconciled view. Variances can be traced to underlying activity, while AI prepares an initial explanation and identifies exceptions needing attention. Finance retains responsibility for validation, judgment and final sign-off.
The same reconciled foundation can support a refreshed forecast combining statistical models, sales conversations, account-level evidence, operating data and finance judgment. Interactive tools can put forecasts, evidence and scenarios into one view, allowing leaders to inspect changes and assess how an adjustment might affect a quarter or year.
Finance teams become tool builders
Friar said OpenAI’s finance staff use ChatGPT Work and Codex to build custom dashboards and tools. OpenAI research cited in the article found that 40% of finance professionals’ specialized AI use involves work outside traditional finance, while 22% involves engineering-related tasks.
One example is a tool built with Codex by a teammate supporting OpenAI’s advertising business who had not coded before. It converts a monthly advertising forecast into weekly and daily plans, accounts for weekdays and holidays, compares forecasts and keeps figures tied to the approved model.
The company also developed IR-GPT, a custom GPT grounded in approved investor-relations materials for diligence questions. The system can produce a first draft in seconds for work that had taken hours or sometimes continued overnight, but the investor-relations team reviews the draft, adds context and checks consistency.
Controls and value measurement remain central
OpenAI argues that faster workflows need explicit governance. CFOs should define data access, permitted actions, approval requirements and escalation points with IT and governance teams. Outputs should connect to reliable sources, and changes to an approved forecast baseline should require finance authorization.
The company recommends measuring AI through operating outcomes rather than seat counts or token consumption. Useful measures can include close cycle time, automatically reconciled transactions, exceptions requiring review, forecast accuracy, refresh frequency and the time needed to produce a scenario. That approach complements OpenAI guidance on managing AI investments on managing AI investments by linking model use to dependable work, review effort and decision quality.
For finance leaders, the practical implication is to begin with a meaningful decision, map its data, approvals and handoffs, and introduce AI where it can analyze, coordinate or complete controlled work. The objective is not to remove finance accountability, but to create more time for judgment and earlier action.