Proserv & PE Insights · Sunday, September 20, 2026

Hi, this is Koko from Koko Knows.

Let's get right into it, because the sharpest signal since we last spoke is coming out of PwC Switzerland, and it's the kind of thing that should make every partner in this audience sit up. PwC Switzerland is halving its bonus pool, and internally they're framing it explicitly as an AI-driven cost response. This was communicated in town-halls, not in writing, which tells you something on its own. But the substance matters more than the delivery method. This is the first hard, on-the-record proof point that AI's margin math isn't staying in the pricing model or the delivery methodology. It's landing directly on paycheques. For years we've talked about AI compressing fees and squeezing realization rates. Now we're watching a Big Four member firm turn productivity gains into cost savings rather than reinvestment in its people. If you're a partner thinking about your own comp pool next cycle, or a PE operator looking at a professional-services portfolio company, this is the leading indicator you've been waiting for. The question isn't whether AI changes the economics anymore. It's who captures the value, the firm or the talent, and PwC Switzerland just answered that question out loud.

The second big move is KPMG teaming up with OpenAI on what they're calling a headless enterprise tech model. This is a meaningfully different bet than what we've seen from most consulting incumbents so far. Instead of bolting AI features onto existing software stacks, KPMG is trying to rebuild the delivery architecture itself around AI-native software, with OpenAI's capabilities exposed more like an API layer than a traditional application. Think of it as consulting infrastructure with the interaction layer ripped out and replaced by an agent. This matters strategically because it's a land grab. KPMG is trying to lock in a platform alliance before its rivals do, and if you've been telling yourself that Big Four firms are slow-moving legacy shops that AI-native vendors can out-innovate, this should recalibrate that view. PwC already picked up a Forrester Leader rating, and now KPMG is moving on architecture, not just messaging. If your growth strategy depends on being the nimble, AI-first alternative to the consulting incumbents, you need real differentiation in price and delivery model now, because the label alone isn't going to hold.

Third, and this one's a governance story more than a headline story: Deloitte surveyed twenty-five thousand UK workers and found that one in three GenAI users are hiding their usage from employers, and half of all users have had no training at all. This has quietly moved from a pilot-stage curiosity into a real billing-risk and client-confidentiality exposure. If a third of your workforce, or your portfolio company's workforce, is using generative AI tools nobody signed off on, you've got shadow AI running through client deliverables, billing narratives, and quality control processes with zero oversight. For PE operators, this is a diligence question you should be asking today, not next quarter. What's the actual usage rate versus the disclosed usage rate in your portfolio companies, and what's the training gap doing to work product quality and audit trails.

Let me tie these together with the implication that matters most for this room. The PwC bonus cut shows the margin pressure is real and it's hitting comp. The KPMG-OpenAI deal shows the incumbents are racing to control the technology layer before it's commoditized. And the Deloitte data shows the risk exposure is already inside the walls, whether firms have acknowledged it or not. If you're a Partner or MD, the strategic move is to get ahead of all three at once: renegotiate how AI-driven productivity gains get shared with your people before someone else's cost-cutting becomes your retention problem, get serious about your own architecture bet rather than waiting to see who wins, and audit your shadow AI exposure before a client or a regulator forces the conversation. For PE, the diligence checklist just got longer. Ask about AI usage rates, ask about vendor breach disclosure terms, and treat agentic security as its own line item, not something buried inside a platform fee. Still running in the background, Accenture's tokenomics research shows that most executives still can't cleanly account for their AI spend, which means there's a wide-open advisory market for whoever moves first on cost governance. And Cognizant continues deepening its Anthropic partnership, which is quietly building a delivery speed advantage that firms without a comparable AI lab relationship are going to find hard to match.

That's the rundown. I'm Koko, and this has been Koko Knows. I'll be back with more as soon as the picture shifts again.