After 25 years advising companies from the garage to the public markets, I have learned to pay attention when a room of CFOs stops being polite. That happened on Tuesday, July 22, at our Foley office in Palo Alto, where we hosted the latest gathering of the CFO Executive Forum. The room stayed full past closing time.
The topic was the one every finance leader hears in every board meeting this year: AI as a force multiplier. How do you scale a lean finance team without adding headcount? We wanted the operator version, not the keynote version. As a deal lawyer, I had my own reason to listen closely. Claims about AI adoption now show up in fundraising decks, diligence requests, and purchase agreements. What gets said on a panel in July gets tested in a data room by December.
The Room
Christina Bui of Robert Half and Protiviti moderated, and she asked the second question, the one behind the polite first answer. The audience was curated. More applicants were turned away than admitted. The goal was a real conversation, not a bigger crowd.
On the panel: Chithra Rajagopalan, Head of Finance at Obsidian Security, who runs a lean finance and revenue operations function inside a fast-growing cybersecurity company. Hull Xu, CFO of Centific, a former investment banker who now runs finance for an AI data company with 3,500 people in eight countries. And Jeff Epstein, Operating Partner at Bessemer Venture Partners and former CFO of Oracle, who works with CFOs across the Bessemer portfolio and serves on the boards of Autodesk and Twilio.
My co-hosts in this community are Christina, Ellie Yashiro of Silicon Valley Bank, Janice Berthold of Heffernan Insurance Brokers, and Murray Newlands of Open Future Forum. We ran the evening the way we always do. Off the record. No pitches. No plus-ones. So I will keep the attribution light where it should be and the lessons intact.
What I Took Away
1. Everyone is using AI. Almost no one has transformed with it. Jeff offered the clearest benchmark of the night. Software engineering is furthest along. By some counts, more than half of new code is now written by machines. Customer support is next. He described a support team that built its own tools, now resolves the easiest 80 percent of tickets without a human, runs a third leaner, and makes fewer mistakes than before. Then he asked the finance question. Has anyone seen revenue per finance employee jump 30 percent? Not a hand went up. Finance teams are saving ten minutes here and a half hour there. That is real, but it is not transformation. The change will come through point solutions, like an AI billing tool he described that reads every customer contract, converts the terms to revenue recognition, and feeds the ERP. Humans review it at first. Within months, it runs nearly on its own. Having reviewed revenue recognition disclosures for two decades, I will believe full automation when the auditors sign it. But the direction is clear, and his bet is that best-of-breed tools beat the big suites for the next decade.
2. The data layer is the whole game. The most practical lesson came from the trenches. Chithra walked us through a real transformation. Her team pulled data out of the CRM, the call intelligence tool, the warehouse, and the product databases, then normalized it into one layer and built dashboards and models on top. The unlock was not the model. It was the normalization. She also learned a hard lesson about context. After feeding an AI assistant every instruction and nuance for pipeline modeling, she came back the next day and it had forgotten everything. Nobody should re-prompt context for a living. You need a layer that holds it. Discipline matters too. At one point, several teams were burning triple the compute on the same problem because nobody had centralized the work. Her formula is simple. Centralize the data. Centralize the definitions. Fund only projects with outcomes tied to company strategy. Any lawyer who has fought over the definition of 鈥渞evenue鈥 in a purchase agreement will tell you she is right. Definitions are where the money is.
3. Global means local. Hull manages teams in eight countries, and his view was a useful reminder that most AI models carry a US context by default. His team could not solve something as basic as expense reimbursement with off-the-shelf tools, because receipts differ by country in format, language, and regulation. So they trained their own model on regional receipts. He builds the culture with creativity too. His team in India ran a two-week hackathon and gave 200 engineering students real problems to solve, including accounts payable. The winners got job offers. The process got automated. His standard for the whole function: if your team is doing the same thing a year later, you are not managing it correctly.
4. Enterprises are saying no to agents, and not without reason. I raised this one myself, because it surprised me in my own practice. Procurement departments at some of the largest companies in the country are refusing to let AI agents run on their networks, and my clients selling agent products are hitting that wall. The panel confirmed it, and the room heard why. In one real incident, an agent with excessive permissions reached into an HR system, changed compensation data, and sent records to a personal email account before it was caught. The fix is not to ban agents forever. The fix is agent identity. Treat an agent like an employee, with defined permissions, an expiration date, and monitoring. From the legal seat, I would add one more thing. Put it in writing. Boards should see an agent governance policy the same way they see a trading policy or a delegation of authority. When something goes wrong, the first question in the room will be who authorized this and under what controls.
5. Your board is benchmarking you, personally. From the board seat, the diligence question is simple. Is this company leaning into AI or holding back? And is the CFO personally leading the effort, or handing it off to a side project? Investors compare your sophistication against every other CFO they work with. Landing well below the bar is a red flag with consequences. The signals of substance over theater are concrete. Forecasts that hit. An analytics function that can answer 鈥渨hat is our revenue per customer in Brazil?鈥 in thirty seconds, not next week. A finance organization that is never the reason the company cannot grow. I sit on the other side of these processes, so I will underline this. In a fundraise, a sale, or an IPO readiness review, your AI story will be tested. Diligence teams have learned to ask the third question. If your AI-enabled finance function is a pilot and a press release, it will show, and it will cost you valuation or credibility. Usually both.
6. The auditors have noticed, and raised their rates. AI lets audit teams read every contract instead of sampling. That is progress. But when finance processes run on tools that do not give the same answer twice, auditors add new procedures to check the machine. Fees go up, not down. The panel expects that tension to fade, but it is real today. On whether standard compliance rules will close the valuation gap between Silicon Valley and everywhere else, the answer was no. The gap was never about compliance. It is about market size. I have closed deals on four continents, and I agree.
Next Up: A Day on the Bay
The CFO Executive Forum is our invitation-only, off-the-record community for senior finance leaders, built with Christina, Ellie, Janice, and Murray, with Foley proud to recommend and support. Give more than you take. Open Future Forum is a global executive community founded in Silicon Valley. Its network reaches tens of thousands of executives and investors worldwide. It runs a year-round calendar of events for senior executives and investors, including CEOs, CFOs, CMOs, CISOs, private equity leaders, founders, and AI leaders, through Forum Select, its invite-only private gatherings, and Forum Events, its open panels and gatherings. Beyond events, Open Future Forum convenes peer groups and executive boards and publishes original research built on first-party survey and qualitative data from its executive network.
Our next gathering is a boat trip on the Bay on August 25. Just a boat, forty finance leaders, and good conversation. Space is limited. More events will follow this fall. Membership is by invitation or referral, the calendar lives on Luma, and you can always reach out to me directly.
My thanks to Chithra, Hull, and Jeff for their honesty, to everyone who filled the room and stayed late, and above all to Murray and Christina for co-steering this forum.
Louis Lehot is a partner at 番茄社区; Lardner LLP, based in Silicon Valley, where he advises entrepreneurs, investors, and public companies on corporate, securities, capital markets, and M&A matters, from garage to global. Views are his own and not legal advice.
