The reason your client never hired PwC was arithmetic, not loyalty. The engagement minimums didn’t work at their size, and the Big Four didn’t want the work at a price your client would pay. That arithmetic just changed.
PwC, EY, KPMG, and Deloitte, the four largest accounting and consulting firms in the world, have collectively invested more than $10 billion in AI since 2023, and the investment landed on their cost structure. The floor under a Big Four engagement came down. Work they used to turn away because it couldn’t carry their overhead now carries it comfortably, which means the 120-person accounting practice and the 75-person law firm they ignored for forty years are, this year, a market.
Between October 2024 and June 2025, all five major firms launched dedicated AI agent platforms: Accenture AI Refinery, Deloitte Zora AI, EY.ai, KPMG Workbench, and PwC’s agent OS.
This comes up in conversations with firm leaders more than any other competitive threat. It’s real and worth understanding clearly. It isn’t the whole story.
What they are doing
EY deployed 150 AI agents across a tax practice of 80,000 professionals, and plans to scale to 100,000 agents, handling the work that junior and mid-level people used to do. At 100,000 agents, EY is writing a staffing plan.
PwC became OpenAI’s largest enterprise customer and has been cutting headcount globally while its cost per engagement drops. Then PwC went further. In early 2026 it launched PwC One, a subscription platform offering automated M&A due diligence, tax guidance, and other services “without a PwC person in the loop.” The client buys an annual subscription and never sees an hourly bill. PwC’s US CEO, Paul Griggs, told partners that anyone who thinks they can opt out of AI is “not going to be here that long.”
That’s a productized offering competing directly with the work mid-market firms do every day. The mid-market reply is the same offering at smaller scale: fractional CFO advisory priced at $3K–$12K/month, with named scope on a one-pager.
KPMG launched Workbench with a target of $12 billion in added revenue. That number doesn’t come from its existing client base alone. The rest comes from serving a market segment the firm couldn’t reach before. In May 2026 KPMG put Claude inside Digital Gateway, the platform where its tax expertise, proprietary tools, and client data already live, and Anthropic named KPMG its preferred consultant for deploying Claude into private equity portfolio companies. KPMG’s vice chair of tax, Rema Serafi, says a regulatory tax agent that used to take weeks to assemble now ships in minutes.
Startups are applying the same pressure from below. Sequoia Capital published a thesis in March 2026 calling professional services the next trillion-dollar opportunity for startups, and Y Combinator followed with the founder playbook. These companies enter a vertical carrying none of the legacy cost structure an incumbent firm carries, so the price they can afford to charge starts below yours. Harvey AI sells lawyers a copilot. The next wave skips the professional entirely: Crosby drafts NDAs for companies directly, Rillet closes books without accounting firms, Mercor places candidates without staffing firms. Their customer is your client.
The mid-market firm faces a pricing environment that’s getting tighter from both directions. But where the Big Four model breaks down is more interesting than where it works.
What they are built for
The Big Four are optimizing for scale. AI agents serving thousands of clients with standardized processes. That’s their advantage and their constraint.
Their model works when the work is repeatable, pattern-driven, and high-volume. Compliance. Standard advisory. Recurring engagements that follow a playbook. AI handles this well at scale, and a global firm with 100,000 AI agents can deliver it at a cost no mid-market firm can match on price alone.
The volume work is what they are coming for.
If your firm’s strategy is to compete on that volume work, on price, against a firm with a $10 billion AI investment, the math doesn’t work. But that’s only one possible strategy.
If the ten-year client still pays by the hour, PwC can beat the price.
What scale does not buy
The Big Four have limits that scale doesn’t fix.
A managing partner at a 60-person accounting firm knows her clients’ businesses in a way that no AI agent or Big Four engagement manager ever will. She sat through the family succession conversations, the cash flow crises, the strategic pivots. She built that context over years. It doesn’t transfer, and it doesn’t replicate at scale.
Twenty years in one regional market teaches a firm which industries actually drive the local economy, how the regulator behaves in practice rather than on paper, and which competitor is about to be sold. None of that is written down anywhere an agent can read it. It’s the context behind the numbers, and a client asking a hard question is usually asking about the context.
Then there’s speed. A client who needs a decision by Friday doesn’t benefit from a global firm’s approval chain. Mid-market firms can scope, staff, and start an engagement in days. The Big Four, even with AI, move at the speed of a large organization.
Customization is the last limit. The Big Four’s economics depend on standardization, and their AI agents work because the processes are repeatable. But the client relationships worth the most are the ones where the firm adapts to the client rather than the reverse. A 50-person firm can build a service around one client’s particular needs. A global firm running agents at scale cannot.
The common thread: clients with complex problems want a person who knows their situation and will give them a straight answer the same day. The Big Four can offer technology, brand credibility, and scale. They can’t offer the partner who already knows the client.
What that leaves you
Knowing your advantages is different from acting on them.
The first thing to move is the mix. Volume compliance work is where the Big Four are headed, and advisory, strategy, and judgment work is where they are weak. A 50-person accounting firm that redesigns its offerings around continuous advisory and proactive risk identification is doing work EY’s agents do not sell.
Pricing follows the mix. Competing on hours means competing on cost, and that fight goes to whoever has the larger AI budget. A client who pays a monthly retainer for ongoing advisory is buying something a compliance engagement never gave them, and the price answers to that rather than to the clock.
The most useful application of AI in a mid-market firm is work the firm couldn’t do at all before: monitoring a client’s financial position in real time, catching the regulatory change that affects their business before they do, reading patterns across the whole portfolio that no single engagement would surface. That’s also what the PE buyers entering your market are asking about. They want to know whether you’ve built anything proprietary, and what your firm is worth increasingly depends on the answer.
A Big Four firm can always offer a lower price. It cannot easily displace a firm that is embedded in how a client operates: in their data flows, their planning cycles, their decisions. A competitor there has to replace your expertise and your accumulated context, and the second one takes years.
Commodity hours, and named advisory
The Big Four are coming downmarket, and ignoring that would be a mistake. But the competitive picture isn’t “bigger firms with bigger budgets will take your clients.”
There are two kinds of work here. One is the commodity volume billed by the hour, and the Big Four are heading straight for it. The other is continuous advisory, which neither the Big Four nor the startups can match.
The managing partner at that 60-person firm sat through the succession fight and the quarter payroll nearly missed. She carries that history into every answer she gives. She has never put a price on that, apart from the hours.