Paul Griggs, U.S. senior partner and CEO of PwC, said the Big Four firm plans to adjust its billing model in certain areas to factor in AI, potentially without even a human professional in the loop, and added that if any of those same human professionals have a problem with it they have no place in this firm.
Griggs, in an interview with the
Grigg, in the interview, appeared to have little patience with those who would hesitate to use AI. He said that senior staff who are not thinking of ways to be AI-first would likely be replaced by others more comfortable with using it, and anyone who thinks they can opt out of AI won't be at the firm long.
"I don't think anyone gets a free pass here. Anyone," he told the FT. The firm is ramping up its hiring of technology experts. "Am I recruiting the same number of accountants and traditional consultants vis-a-vis engineers, on a proportionate basis, that I was three years ago? No," said Griggs.
A specific example of the firm's AI technology is the PwC One platform, which
It's meant to let clients interact with AI-enabled capabilities in real time while work continues to be guided by PwC's standards and expertise. It would also be used directly by PwC professionals, with the platform serving as a sort of "digital front door." This is all with the intention of moving beyond episodic projects toward more continuous insight, earlier identification of emerging risks and opportunities, and deeper collaboration between PwC professionals and their clients.
Built with agentic AI, it has a variety of use cases that include acting as a financial statement analyzer that reviews client financial statements against PwC's tax playbooks to highlight potential areas worth exploring; a tax incentives explorer that identifies a broad range of potential tax incentives, grants, credits and other funding opportunities based on client operations; a transfer pricing policy compliance reviewer that compares client transaction data to documented policy to identify potential misalignment; an intercompany agreement reviewer that reviews client intercompany agreements against actual transactions to help identify inconsistencies early; a greenhouse gas emissions navigator that uses anomaly detection solution to help organizations proactively identify unusual patterns, data quality issues and emerging risks across sustainability data; or a financial due diligence assistant that provides an initial view of key financial and related deal risk, accelerating early-stage decision-making.
Likens said these capacities are best understood as agent-driven use cases applied to specific client needs and workflows. Additional use cases will continue rolling out over time as the platform expands into new areas.
The platform itself is built on a multi-agent architecture that combines traditional AI, generative AI, machine learning and other techniques to orchestrate specialized capabilities that work together, from surfacing patterns and testing assumptions to accelerating analysis within PwC's methodologies and workflows. It is accessed via a conversational interface so users can engage in plain language, ask questions and guide the system much like they would in a chat. Beyond the chat interface, it also features a dynamic, structured canvas that surfaces data, workflows and outputs in real time. Likens said that as the user interacts, whether by prompting or selecting elements, the canvas updates to reflect analysis, assumptions and next steps.
"Rather than clicking through static menus, it's a more fluid experience that combines conversation with interactive, context-aware visuals and tools," he said.
While clients can make use of these capacities on their own, Likens said a PwC professional remains in the loop in every case — reviewing outputs, applying judgment and maintaining alignment with PwC standards, regulatory requirements and the client's specific context.
"That human oversight is a critical part of how the platform is used today, helping maintain the level of trust, quality and accountability clients expect," he said.
While Griggs implied this may not be strictly necessary, as some of the firm's AI-powered automated tools might potentially be accessed "without a PwC person in the loop," a spokesperson later clarified that he was talking about only initial access. Depending on the specific use cases clients are interested in exploring, as well as next steps after initial use, there would be PwC professionals involved in follow-up engagements.
PwC One's pricing model will evolve over time, but may involve subscription or consumption-based pricing elements.
