AI productivity – Anthropic CFO Krishna Rao says AI is boosting productivity by taking over parts of white-collar work, changing roles rather than eliminating workers.
AI is already writing a large share of tasks once handled by people—and Anthropic’s leadership says that shift is redefining what “work” looks like, not simply shrinking headcount.
In remarks on an episode of the “Invest Like the Best” podcast published Wednesday. Anthropic CFO Krishna Rao described a workplace where AI increasingly performs the execution layer of knowledge work.. He pointed to areas ranging from software engineering to financial reporting. arguing that humans are moving toward oversight. judgment. and strategy as the day-to-day mechanics become more automated.
Rao said the company is hiring more people as AI boosts productivity.. “We’ve hired a lot more people because of that. ” he said. framing AI as a productivity accelerant rather than a substitute for work.. In his view. teams can ramp faster on how to use Anthropic’s tools. which makes new hires more productive as they learn the company’s workflows.
Inside Anthropic, the coding portion of work is already heavily delegated to AI systems.. Rao said that “90 plus percent” of the company’s code is written by its Claude Code product.. The implication is that engineering output is being produced with less manual drafting by developers. while human efforts can shift toward reviewing. directing. and making higher-level technical decisions.
The finance function is also being reshaped.. Rao said Anthropic uses Claude to produce financial statements. and that its monthly financial review process is described as “90 to 95% ready” before humans step in.. That means reviewers are no longer starting from scratch. but instead evaluating AI-generated outputs and interpreting what the numbers indicate for the business.
Rao added that some internal reports that used to take hours to assemble are now reduced to roughly 30 minutes in preparation.. He connected the change to a reallocation of time: fewer hours spent gathering and formatting information. and more attention devoted to decision-making.. For a finance team, that distinction matters because it shifts value from producing artifacts to assessing implications.
Beyond the specific examples at Anthropic, Rao argued that AI can amplify the talent a company already has.. He described a “talent density” approach—seeking concentrated expertise in areas such as AI research talent and inference engineering talent—supported by what he said are strong models.. The message is that AI can make highly specialized teams more effective, rather than replacing them.
The broader debate over AI and jobs is intensifying across industries. and Rao’s comments place Anthropic’s stance within that discussion.. He said some companies have urged employees to adopt AI as automation improves. while others have cut jobs while citing efficiency gains from AI-driven systems.. That contrast underscores why economists and executives disagree on what will happen next: whether automation reduces work faster than new roles are created. or whether productivity gains expand labor demand.
Rao suggested Anthropic’s internal experience aligns with the more expansion-oriented view.. As employees become more productive with AI assistance. companies may expand hiring because “there’s no shortage of work to do. ” he said.. In other words, if AI increases throughput, organizations may find additional projects that need human judgment and leadership.
Still, the most striking part of the shift Rao described is qualitative: the nature of white-collar work is changing.. He said employees increasingly oversee AI systems rather than completing tasks manually. and that teams can deploy “fleets of agents” working across projects at the same time.. That structure points to a managerial model built around directing tools and monitoring outcomes rather than doing every step by hand.
Rao summarized the workplace change with a blunt workplace reality—“Everyone kind of becomes a manager.” He framed the productivity gain from that model as early but potentially significant. describing what he sees as enormous potential in using AI systems to expand execution capacity while keeping humans responsible for direction and judgment.
For employers watching the market, the core question becomes how companies design work and evaluate performance.. If AI is handling the production portion of tasks. then internal training. review standards. and decision rights may become as important as hiring. because the job shifts from execution to verification. prioritization. and strategic coordination—skills that are harder to automate than drafts. reports. and first-pass outputs.
The labor implications may therefore hinge less on whether AI can perform tasks at all. and more on how quickly organizations adjust their operating models.. In Anthropic’s case as described by Rao. automation appears to be integrated into both engineering and finance workflows. while the company’s growth approach remains tied to productivity—supporting the argument that AI can change roles and responsibilities even as companies continue to scale their teams.
Anthropic Krishna Rao Claude Code AI productivity white-collar jobs financial reporting automation
So basically the robots are doing the office work now? cool cool.
I don’t buy it. If AI is writing 90% of the code and “hiring more people” then where are the jobs actually? sounds like corporate spin to me.
“Oversight and strategy” yeah okay but who pays for all that oversight? Like the CFO said people are moving up but it’s still the same paycheck level for most folks. Also 90 to 95% ready financials… that seems risky? what if the AI is wrong and nobody catches it fast.
I saw something like this and it always ends up being layoffs later. First they “boost productivity,” then they stop hiring, then suddenly your team is gone. And 90+ percent of code being written by some tool… are they even testing stuff or is it just generating whatever looks right? anyway I bet most people are just gonna be managing prompts now.