Sarah Friar joined OpenAI two years ago. The books still closed the old way. Manual hunts for records. Endless spreadsheet stitching. Late-night explanations of variances. Forecasts assembled from static snapshots.
That changed. Not with incremental automation. With a full rethink of how finance should work when the most powerful models sit at every desk. Friar laid it out this week in a detailed post on the company’s own site. The goals sound almost absurdly ambitious. A zero-day close. Continuously refreshed forecasts that update without waiting for month-end.
OpenAI’s own blog post makes the case plainly. “We are still building toward both ambitions,” Friar wrote. “The work has already changed how our team operates. It has pushed us beyond the limitations of static spreadsheets, manual searches for supporting records and presentations toward live tools built on the full context and data of the business.”
Short sentence. Big claim. Yet CFOs across industries now watch closely. Because OpenAI operates at the frontier of both AI capability and eye-watering compute costs. Its own financials reflect the tension. The company reported a net loss of $38.5 billion in 2025 on $13.07 billion in revenue, according to Yahoo Finance coverage from late July. Growth remains ferocious. Annualized recurring revenue in July alone topped the entire second quarter, Friar told employees in an internal meeting reported by CNBC.
But scale brings complexity. Billions in GPU spend. Rapid hiring. Investor scrutiny ahead of a potential public listing. Friar has said she no longer reports directly to CEO Sam Altman. She has questioned spending levels. The tension between speed and control sits at the heart of her finance overhaul.
The zero-day close concept isn’t new. Finance technology vendors have pitched versions of it for years. What OpenAI attempts differs. The company wants a reconciled, traceable view of its position available in real time. Approved spending plans linked to general-ledger actuals, purchase orders, accruals and transaction details. All continuously reconciled. No frantic scramble when the calendar flips.
“The idea behind a zero-day close is to give leaders a real-time, reconciled and traceable view of the company’s financial position,” Friar explained in the post. The broader aim feels even more strategic. A finance team that “understands what is happening as it happens, shows leaders what could happen next and helps the company make better decisions sooner.”
Easy to say. Harder to build. Friar stresses that success demands more than plugging in new models. Teams must redesign workflows around the actual decisions that matter. They must experiment freely yet maintain accountability. And they must measure the value produced by AI with precision. Her recently published scorecard for AI economics, covered in CFO.com on July 17, drives exactly this point. The core question for every CFO: Does the value of work completed by AI grow faster than the cost of producing it?
OpenAI’s internal experiments already deliver results. Every finance team member now builds custom dashboards and tools using ChatGPT and Codex, the company’s AI coding agent. One colleague with zero prior coding experience created a system that converts monthly advertising forecasts into weekly and daily versions. The output? Faster course corrections. Less guesswork.
Investor relations gained its own specialized tool. IR-GPT draws only from approved materials. What once required hours or overnight effort now produces a strong first draft in seconds. Human reviewers still check sources and add judgment. The speed gain compounds.
Friar shared five lessons from the effort. Give everyone access to the tools. Redesign entire workflows around decisions rather than tasks. Turn finance professionals into builders. Pair acceleration with strict accountability. Measure value per unit of intelligence produced. Recent OpenAI research shows 40% of specialized AI usage by finance professionals now involves work outside traditional accounting. Another 22% touches engineering-related tasks. The lines blur.
She has grown blunt about hiring. “I would never hire a finance person who didn’t know how to use Excel, and I probably wouldn’t hire a finance person today that doesn’t know how to use a tool like Codex,” Friar said at the Liquidity Summit, as reported by The Times of India in June. The comparison lands. Spreadsheets once separated professionals from clerks. AI fluency now does the same.
Other companies experiment in parallel. Startups such as Rillet and Numeric promote AI-driven close management that targets the same real-time ideal. A March 2025 analysis from Numeric called the zero-day close increasingly likely thanks to better data pipelines and intelligent automation. Yet most organizations still treat it as aspirational. OpenAI’s advantage lies in owning the models and the data flywheel. Its finance team can train systems on internal context at a depth few rivals match.
Still, limits remain. Friar herself noted in recent remarks that AI speeds finance work but cannot replace human judgment on complex accounting questions or strategic trade-offs. GPU costs continue to strain margins despite revenue growth, she has acknowledged in multiple forums.
The potential IPO adds pressure. Confidential S-1 filing completed in June. Public markets will demand transparency and controls far stricter than a private startup requires. A finance function already operating in near real time could prove a powerful advantage during the transition. Or it could expose weaknesses if the automated forecasts miss material shifts.
Friar’s approach rejects the common trap of simply speeding up yesterday’s processes. “We had access to the most advanced AI tools in the world, and we were still learning how to redesign finance around them,” she reflected. That redesign focuses on decision velocity. Connect the data once. Reconcile continuously. Surface insights before leaders ask. Let humans focus on interpretation, negotiation and foresight.
Industry observers see echoes in other high-growth technology firms. Yet few possess OpenAI’s combination of internal model access and public visibility. The company’s experiments therefore carry outsized weight. When the CFO of the organization many view as the AI leader publishes lessons on overhauling her own department, peers take notes.
Results so far appear promising but incomplete. “We are still building toward both ambitions,” Friar repeated. The phrase carries honesty. Zero-day close and perpetual forecasting remain works in progress. The shift in daily operations already feels tangible. Less time hunting data. More time analyzing outcomes. Custom tools that adapt as the business evolves.
And the implications stretch beyond one company. Finance leaders everywhere now face the same basic economic test Friar outlined in her AI scorecard. If intelligence becomes cheaper and faster, the value created must outpace the expense. Otherwise the investment collapses under its own weight.
OpenAI’s bet looks clear. Build the systems that make financial truth available instantly. Arm decision-makers with forecasts that update by the hour. Turn accountants into tool creators. Accept that the close never really ends. It simply becomes part of the continuous flow of the business.
Whether that model survives the scrutiny of public markets or the next wave of model costs remains uncertain. But the direction Friar has set for her team already influences how other CFOs think about their own transformations. Real-time. Accountable. Built by the people who will use it. No more waiting for the books to close.