The signals that investors relied on to identify great founders in the first AI/ML wave are already shifting. The firms that update their pattern recognition fastest will capture the best companies of this wave.
At our Scope 2026 conference last month, I sat on a panel with Phil Petitpont (CEO of Moments Lab, an Oxx portfolio company), Michelle Coventry (VP of Talent at Creandum), and Ivan Levingston of the Financial Times to discuss what a great founder looks like in 2026. I had just published our whitepaper studying 645 founders across 270 software companies spanning the SaaS era through to today’s AI application layer. The panel gave me the chance to test the findings against two people who see founders from different vantage points: Phil, who is building an AI company and hiring a team in this environment right now, and Michelle, who works across a hundred portfolio companies and tracks founder talent globally.
Three things from that conversation have stayed with me.
Intellectual honesty has overtaken technical credentials.
When I shared the finding that the PhD premium is declining (18% in the first AI/ML wave, already down to 12% among the newest companies), Michelle’s response was immediate. The dial that has gone up at Creandum, she said, is not about technical depth. It is about intellectual honesty. “When was the last time you changed your mind? Show me what you’ve built. It’s not about being a technical founder. It’s about being deeply curious enough to go and discover the answers for yourself.”
That is a meaningful shift. In the first AI/ML wave, the question investors asked was “can this person build the technology?” Now, with models commoditising and AI development tools collapsing the barrier to building, the question is becoming “can this person see the problem clearly, build the right product around it, and adapt when the ground shifts?”
Phil reinforced this from the operator’s side. At Moments Lab, he has restructured the entire product organisation. No more frontend developers, backend developers, and product managers as separate roles. Instead, three-person teams, each with a single objective, empowered to reach it however they choose. “We are looking for people obsessed about delivering value,” he said. “Not technical experts.” When I asked about the shift in his own leadership, he described hiring an AI chief of staff agent, finding that it matched the quality of advice he was getting from experienced human advisors, and using it to reduce his cognitive load so he could focus on strategic decisions.
The practical implication is clear. If the best companies are being built by people who are obsessively curious, fast to adapt, and willing to unlearn what worked before, then investors who screen primarily on credentials are selecting for the wrong signal.
The serial entrepreneur bias is no longer supported by the data.
A received wisdom in venture capital, backed by data from past eras, is that previously successful serial entrepreneurs are the best predictor of venture success. Our whitepaper suggests that is no longer the case. Among successful AI/ML wave founders, 60% had never founded a company before, compared with 50% for SaaS. The first-time founder rate shifted because the technology itself selected for people who had specific technical expertise, not commercial track records.
Michelle’s observation on this was sharp. She described working with both first-time and repeat founders and noted that repeat founders often need to “unlearn” as much as first-time founders need to learn. “You’ve got battle scars, tissue that needs breaking down,” she said. What people want from companies today is different from when repeat founders were building previously.
For investors, and particularly for LPs evaluating fund managers, this matters. A firm that still weights serial entrepreneur status as a primary signal is using a heuristic that the data no longer supports. The best founder for a category-defining AI company may be someone who has never founded anything before but who has the intellectual honesty, technical curiosity, and adaptability that this compressed cycle demands.
The founder matters more, not less. But so does what surrounds them.
Ivan asked me directly: has the founder become more or less important in the AI era? I said more important. Probably even more important than before. When a founder can develop genuine superpowers through AI tools (in the way Phil described), the gap between a strong founder and a replacement widens. There are two dimensions of ownership in a company: financial ownership, which investors understand well, and emotional ownership, a kind of parenthood. When a founder leaves, the company is orphaned and that dimension of value disappears.
But here is the tension. The founder matters more, and yet founding teams are getting simpler. Our whitepaper shows teams shrinking (2.6 mean in the first wave, 2.3 in the newest cohort), solo founding rising, and the PhD rate dropping. If the founding team carries fewer of the capabilities a company needs to scale, those capabilities have to come from somewhere: the first commercial hire, the board, the advisory network, the investor.
Michelle captured this well. She noted that in the last 18 months, there have been more requests from founders for support, coaching, and private discussions than in any period before. “The best founders don’t need VC help on people and talent” used to be the line. That has changed. Everything is happening so fast that even the strongest founders are asking for help building the teams around them.
Phil described working closely with Oxx on hiring a CFO, a process he said produced one of the best hires in the company. That is not a founder who did not need help. That is a founder confident enough to say “I need help here” and a board relationship healthy enough to provide it. At Oxx, we call this vulnerability-based trust, and it is one of the most reliable indicators of organisational health.
The so what, for investors and LPs.
The pattern recognition that served investors well in the first AI/ML wave (PhD founders, serial entrepreneurs, research lab pedigree) is becoming less predictive. The signals that matter now are harder to screen for: intellectual honesty, speed of adaptation, the ability to build with AI tools, the self-awareness to ask for help. These are not signals you find on a CV. They emerge in conversation, in diligence, and in how a founder responds when the ground shifts underneath them.
For LPs, the question is whether your fund managers are updating fast enough. As Bob Thomas put it in closing Scope: “Most of us in this room are professionally good at pattern recognition. The patterns are changing, and we’re having to re-evaluate our priors on good founder composition, good team composition, good execution.” The firms that do this best will back the companies that define the next wave. The ones that do not will keep pattern-matching on a world that has already moved on.
This piece reflects a panel discussion at Scope 2026 featuring Michelle Coventry (Creandum), Phil Petitpont (Moments Lab), and Richard Anton (Oxx), moderated by Ivan Levingston (Financial Times). It draws on Oxx’s whitepaper covering 645 founders across 270 companies.
Read the full whitepaper: Who will build the next wave of breakout companies?
