On March 19, PwC launched PwC One — an AI platform offering six automated services on a subscription. Financial statement analysis against PwC's tax playbooks. Tax incentive identification. Transfer pricing compliance. Due diligence. GHG emissions anomaly detection. More coming monthly. Clients log in, describe a problem, and autonomous agents do the work while PwC professionals review in the background.
The same week, PwC's US CEO told partners that anyone who resists AI "has no place at this firm." Those aren't separate stories. They're the same story.
The hourly model just got a shelf life
PwC isn't experimenting. They're restructuring. Bonuses now track revenue and margin per professional, not hours billed. Leaders are measured on progress moving services to AI. The firm is hiring fewer traditional accountants and more engineers and data specialists. The entire incentive structure has pivoted from "bill more hours" to "deliver more value with fewer hours."
The pricing signal is already showing up in client conversations. Bloomberg reported that PwC clients have demanded AI discounts — arguing that if AI makes the work faster, the savings should flow to them. PwC's response wasn't to defend hourly rates. It was to cut prices where AI saves staff time and shift toward subscription and consumption-based pricing.
When the largest professional services firm on the planet decides that hourly billing is a losing strategy, that decision doesn't stay inside PwC. It resets client expectations across the entire market. Your clients read the same headlines. They'll start asking the same question: if AI is doing the work faster, why isn't my bill lower?
What PwC can encode — and what it can't
PwC One automates what Sequoia Capital's intelligence-versus-judgment framework calls the intelligence layer — the rules-based, pattern-matching work that practitioners have spent careers internalizing. We hosted a webinar on the Sequoia manifesto just over a week ago — if you missed it, view the full session and notes at theaiaccountant.ai/blog/sequoia-webinar. PwC is now operationalizing that thesis at scale.
Financial statement analysis against standardized playbooks? Intelligence. Tax incentive identification? Intelligence. Transfer pricing compliance? Intelligence. Patterns — learnable, encodable, automatable. PwC encoded decades of institutional knowledge into a multi-agent platform. That compresses the value of the intelligence work that smaller firms sell.
But PwC One can't do what you do on a Thursday night when your client calls because their business partner just emptied the operating account. It can't read the room during a meeting and know that the real question isn't the one being asked. It can't tell a business owner they're lying to themselves about their margins — and have them listen, because you've earned that trust over years.
The relationship. The contextual judgment. The wringable neck — someone's name on the file, someone who knows the family dynamics behind the business decisions. That's judgment. It doesn't live in a platform.
The relationship alone isn't enough
Here's where I need to be direct. "We have better relationships" is not a strategy if your delivery model is slower, less data-rich, and more expensive than what a subscription platform offers. The relationship is the moat — but only if it's paired with delivery that justifies the premium.
PwC's partner rates — $500 to $800 an hour — were the moat that kept them out of the mid-market. A subscription platform has no such constraint. If PwC One matures, mid-market companies who couldn't afford PwC's consulting now access PwC's institutional knowledge at a fraction of the cost. That competes directly with the advisory services you sell.
The answer isn't to pretend the relationship makes you immune. It's to make the relationship worth more by making everything around it faster and smarter.
Encode your knowledge — not for PwC's benefit, for yours
PwC encoded their institutional knowledge into a platform. You need to encode yours — into AI systems you control. SOPs that tell AI how your firm handles specific client scenarios. Client profiles that give AI the context it needs to draft communications, flag anomalies, and prep for meetings. Pricing logic, engagement workflows, advisory frameworks — the institutional knowledge that currently lives in your partners' heads and your senior bookkeeper's muscle memory.
This is context engineering. Not because it's trendy. Because it's the only way to compress the intelligence work so you can spend more time on the judgment and relationship work that justifies your fees. The partner's highest-value mode isn't doing the analysis — it's directing and reviewing the analysis. PwC just operationalized that at scale. You can do the same at yours.
The delivery model that wins
The firms that thrive through this won't be the ones with the best relationships or the best AI. They'll be the ones that build both. Encoded firm knowledge that handles the intelligence work at speed and low cost. Deep client relationships where judgment, accountability, and trust create value no platform can replicate. The intelligence layer runs on your context. The relationship layer gets more of your attention, not less.
PwC just showed you the future delivery model — fewer humans doing production work, more humans doing judgment work, all of it priced by outcome rather than by hour. You don't need PwC's budget to build your version. But you do need to start encoding. The firms that build both own their pricing. The firms that build neither will have it set for them.
Tomorrow, we're moving from "what PwC did" to "how to do it for yourself." Because context engineering isn't a nice-to-have. It's the mechanics of delivery when AI handles the intelligence and you handle the judgment.
Ready to build the delivery model that wins. The AI Practice Transformation program walks you through the full shift: how to encode your firm's knowledge, redesign workflows around what AI makes possible, and price by outcome instead of by hour. Three weeks. Five-hour sessions. Live cohorts with other CAS leaders navigating the same pivot. You're not doing this alone.