Five PE and tech leaders debated AI’s speed, governance, and due diligence at SuperReturn US West. Here’s what actually mattered.
The panel, for the record: Ian Gutwinski (CEO, Mosaic) moderated a conversation with Josh Klinefelter (Partner, Aurora Capital Partners), Lou Rassey (Co-CEO & Managing Partner, McNally Capital), Ben Collins (Senior Director of Solutions Marketing, AlphaSense), and Paul Sloan Zimmerman (Managing Director & Global Head of Private Equity, Google) at SuperReturn US West’s “Redefining Private Equity Success Through AI Integration” keynote panel.
Four private equity operators walked into a conference room forty-eight hours during the same week as the All-In Summit and were asked, essentially, whether the industry building the future should be allowed to keep building it at this speed. Nobody flinched. What followed was a genuinely useful hour on how AI is actually restructuring mid-market private equity.
The Governance Question Everyone Agrees On and Nobody Has Solved
Zimmerman opened with the closest thing to a consensus position in the room: the safety conversation has gotten less attention than it deserves, mostly because the labs moving fastest are also the ones demonstrating where the risk lives. That’s a fair point dressed up as a talking point. What made it interesting was what came next. Nobody on that stage argued for slowing down. Rassey said outright that shaving the pace back fifty or seventy percent wouldn’t change much, because the disruptive force is already large enough at current capability.
Where the Money Actually Is: Due Diligence, Not the Headline Use Case
Collins gave the most concrete example of the day, describing a conversation with an AlphaSense client whose team is now running thousands of virtual data room documents through AI-assisted review, using expert-call synthesis to stress-test a thesis before committee. The detail worth sitting with wasn’t the speed. It was his observation that AI tooling is structurally better at finding reasons to kill a deal (customer concentration, narrative inconsistency, buried liabilities) than it is at building conviction to do one. Conviction still requires a human. That’s not a hedge. That’s the actual shape of the technology right now, and it’s a more honest claim than most of what gets said about AI and deal-making in public.
The Line That Should Worry Every Firm Without an AI Strategy
Klinefelter and Rassey both landed on the same word independently: table stakes. Not a differentiator. Not a competitive edge. The baseline cost of staying in business. Klinefelter’s firm has been running a fifteen-month adoption push across its portfolio with an outside partner (a McKinsey and QuantumBlack spinout called Woven Blade) specifically because internal teams didn’t have the horsepower to diagnose where AI could move the needle inside a commercial engine. Buying the tools is the easy part. Building an internal operating model for using them is the actual competitive terrain, and most firms are still shopping for tools.
An AI Investment Committee Member, Non-Voting, Always On
The most structurally interesting idea in the room came from Rassey: a non-voting AI participant in the IC process, feeding independent, always-on pressure-testing into deal evaluation before analysts even get to the formal meeting. He was candid that the harder problem isn’t building the tool. It’s the human half: getting a committee to actually respect an AI’s dissent when it disagrees with the room’s instinct. Zimmerman’s addition mattered here too. He argued governance built into the deal process from the start beats governance bolted on after the fact, every time, and that the sign-off checkpoint has to exist before AI-generated data reaches a decision-maker, not after.
More detail on the conference session is available via SuperReturn US West’s event page.
Mini FAQ
Is AI actually replacing due diligence analysts at private equity firms?
No, and none of the panelists claimed otherwise. The consistent theme was AI accelerating pattern-matching across large document sets while leaving relationship-driven judgment (the part that actually builds conviction on a deal) to humans.
What does “table stakes” mean in practice for a mid-market GP that hasn’t started?
It means the operational advantage of early AI adoption compounds, per Klinefelter, and firms waiting to see how it shakes out are already behind rather than being cautious.
Justifying Moving Fast
Nobody on that panel resolved the tension between moving fast and building the governance to justify it. What they did agree on, unanimously and without much hedging, is that the firms treating this as a procurement decision are going to lose to the firms treating it as an operating model rebuild.
