OpenAI’s CFO names the new hiring floor: fluency with AI coding tools.
What the consulting field took six issues to name, Sarah Friar compressed into a single hiring filter.
Start with the sentence that moved the line. In an interview on June 5, 2026, OpenAI’s CFO Sarah Friar said she would never hire a finance person who didn’t know how to use Excel, and that she probably wouldn’t hire a finance person today who didn’t know how to use a tool like Codex. That is not a statement about how OpenAI hires. It is a statement about where the floor now sits.
Think about what Excel did when it became the professional standard. Knowing it never made anyone exceptional. Before that moment, the skill got you the job and the people without it were filtered out. After it, everyone who mattered had it, so it stopped being a differentiator and became the baseline. You were not remarkable for knowing Excel. You were disqualified for not knowing it. Friar is saying that exact moment has now arrived for AI tools in finance. Not in some future year. Now. The hiring filter has been updated while most organizations were still treating AI fluency as optional training.
Her sentence is the clearest single version of something these firms had been circling for six runs in a row. Accenture’s Julie Sweet made AI fluency a promotion condition in May, telling people that getting promoted now means doing the things the firm does to operate. BCG’s Sylvain Duranton released the first hard workforce study in the field in June, 10,000 workers across companies, finding that business value and employee enjoyment rise together when the AI strategy is clear, and that people do not resist AI when the direction is real. Cognizant’s Ravi Kumar S named the closest structural version, describing managers turning into player-coaches who can both execute and verify what agents produce, and he rejected the displacement story outright, calling AI an amplifier of human potential rather than a displacement strategy. Bain’s Christophe de Vusser ran a firmwide vibe-coding hackathon as in-house fluency infrastructure and said net job creation outpaces destruction.
BCG’s Christoph Schweizer gave the sharpest commercial reading of it. The new hires, he said, are so AI native that they use these tools as if the world had never existed without them, and so, perhaps counterintuitively, the firm is still hiring. BCG is not replacing people with AI. It is replacing people who aren’t AI-fluent with people who are. The floor moved and the headcount held, and those are not contradictions. They are the same statement. There is, however, a real caveat, and it is the most important point in the run. ISG’s Stanton Jones, the first skeptic in the field, granted that the floor has moved but argued that moving it does not automatically deliver the outcomes being claimed. A finance team can become AI-fluent, but if the organization never redesigns its workflows, the fluency produces nothing measurable. The floor is a necessary condition, not a sufficient one.
Friar also disclosed something the field has not yet absorbed. Even after aggressive pre-buying of capacity, OpenAI will not have enough computing power in 2026. Compute, she said, is a very scarce resource right now, and even with the pre-buy, 2026 still won’t have enough. That opens a gap between the hiring signal and the deployment reality: companies are now screening for Codex fluency in a market where the engine to support that fluency at scale is constrained. The floor moved, and the ceiling did not rise to meet it. Kumar pushed the frontier further in the same window, calling physical AI the iPhone moment for robotics as vision sensors, precise positioning, and low-cost actuators arrive together. The floor Friar named is the software floor. The physical floor has not been named yet, but the field is pointing straight at it.
Excel moved from differentiator to baseline over a decade, because Microsoft made it cheap and everywhere, schools taught it, and every employer required it. The responsibility was spread across vendors, educators, and employers over ten years. AI fluency is moving in years, not decades, and the floor is rising faster than any of those institutions can update their courses, their training, or their job descriptions. The firms on the right side of this have already defined their own floor, not the vague version of knowing how to use AI but specific tools, specific use cases, specific demonstrated capability, and they have built the training to carry their people across it before the floor becomes disqualifying. The firms that have not done this are hiring below the floor right now. They just haven’t gotten the rejection letter yet.
Excel never made you exceptional once it became the standard. It made you qualified. AI fluency has just crossed the same line, and the firms still treating it as optional are hiring below the floor without knowing it.