Quick summary:
- North Korean state actors are using AI-powered deepfakes to pass job interviews at American tech companies, and most hiring processes can’t catch them
- Enterprise AI adoption is a measurement problem, and companies that find their internal “trailblazers” can model successful behavior across the workforce
- A $1.5 billion VC fund says the AI startup bubble isn’t in the technology itself but in the sameness of the pitches, where every successful company has 20 undifferentiated clones
Andrew McLeod runs background checks on job candidates for a living. Lately, some of those candidates don't exist.
David Schulhof flew to China twice last year and saw 150 robotics companies moving faster than anything in the U.S.
Jim Larrison can tell you exactly which employees at your company are using AI, which tools they're using, and whether any of it is producing results.
Ross Fubini manages $1.5 billion in venture capital. He walked an AI trade show floor and got the same pitch from a dozen companies before lunch.
Each came to HumanX with a different vantage point on AI’s impact, and together their conversations paint a picture of an industry where companies are either producing real results or burning cash on tools nobody uses.
They joined host Greg Matusky for a special edition of The Disruption Is Now recorded from the show floor in San Francisco. The guests include:
- Andrew McLeod, CEO and Co-Founder, Certn
- Jim Larrison, Co-Founder and President, Larridin
- David Schulhof, Founder and CEO, Toborlife AI (Unitree Robotics partner)
- Ross Fubini, Founder and Managing Partner, XYZ Venture Capital
Watch now:
Key takeaways
AI-powered identity fraud is infiltrating American hiring
McLeod runs Certn, a global background check and identity verification company with 400 employees. State actors — many of them North Korean — are buying stolen Social Security numbers, names, and address histories from the dark web, then applying for jobs at American tech companies under those identities.
When they land an interview, they show up as an AI-generated avatar with no lag and no tells. They carry LinkedIn profiles loaded with engagement. Hiring managers walk away convinced they've found a strong candidate.
His proposed fix is to flip the hiring model. Instead of screening after the offer, McLeod envisions candidates arriving with verified digital credentials — issued by universities, employers, and professional organizations — so identity is confirmed before the first interview.
China is winning the robotics hardware race while the U.S. leads on software
Schulhof founded Toborlife AI as a U.S. partner for Unitree, the Chinese robotics company known for its humanoid robots and robotic dogs. After two trips to China in the past year, he says the country has roughly 150 robotics companies mass-producing at a pace that echoes the early automobile industry before consolidation.
Toborlife imports Unitree's hardware and builds its own autonomous control software on top of it. The initial market has been universities and large corporations buying robots for developer training and simulation. The shift now is toward real-world deployment. Unitree's robotic dogs are already doing security patrols and facility inspections, programmed to identify threats and alert operators.
Schulhof compared the trajectory to electricity. The robots are primitive now and limited in their capacities, but the acceleration will be fast.
The employees getting the most out of AI aren't telling anyone
Larrison's company Larridin tracks AI adoption by measuring what tools employees actually use across the enterprise. The company's database covers roughly 120,000 AI applications and can show which departments are getting results and which are running up costs with no measurable return.
Larrison keeps finding that every company has a small group of AI trailblazers producing outsized business outcomes, and they have no incentive to share their methods. One customer dug into its sales data and discovered that its highest-performing reps were all heavy AI users. They sent AI-generated follow-up notes after calls. Prospects responded faster, and close rates climbed.
Matusky added his own perspective. His firm has 132 employees, and the tools and training are the easy part of AI adoption. Getting real behavior change from people who have done their jobs the same way for years is the real challenge.
The AI bubble is in the ‘sameness’
Fubini started XYZ Venture Capital in 2017 and now manages about $1.5 billion. He invests at the earliest stages and concentrates capital into his best bets.
At HumanX, he found companies using identical language to describe identical products. "Basically, every single successful company has 20 clones that are tackling it," he said. He got pitched by two agent-workflow startups just walking between sessions.
Fubini doesn't think the demand is fake. Every chip that can be manufactured is in use, and he sees that as real. But he thinks differentiation is what separates the companies that survive from the ones that wash out. You either find a specific context nobody else is solving for, or do something technically hard enough that copycats can't follow.
White-collar jobs will change more than the people doing them can grasp
When asked what's been underhyped, Fubini pointed to the coming transformation of white-collar work.
He believes financial analysts, lawyers, and doctors will see their daily routines reshaped so profoundly that people in those jobs today can't picture it. His own team already uses agents that handle all the background research before anyone shows up at the office. The human work has shifted to real-time conversation such as talking to founders, checking references, and reading people in person.
Fubini thinks those high-earning professionals will be fine. The harder question is what happens further down the income ladder, where fear-driven narratives about technology tend to hit hardest. He worries about workers with the least access to power and money bearing the brunt of the transition while the people with the most resources adapt.
Key moments:
- How state actors build fake identities to pass American background checks (4:56)
- Why some deepfake job candidates are indistinguishable from real people on video (6:51)
- Using verified digital credentials that flip the hiring model (10:38)
- China's 150 robotics companies and the hardware manufacturing gap (14:18)
- The comparison between robotics and electricity (18:07)
- The story of “Glenn,” a sales agent that produced the best first call of a rep's career (27:20)
- The CFO-CIO tension over AI spending (22:33)
- Fubini's surfer-surfboard-wave analogy for evaluating startups (34:38)
- Why every successful AI company has 20 clones tackling it that won’t survive (38:43)
- Making the case against humanoid consumer robots (43:49)
- Why Fubini believes the world bends toward the path of the angels (48:37)
Q&A
Q: Where is the best AI talent coming from globally?
McLeod: In terms of the execution and practical use of AI, India, Pakistan, Nigeria, a whole bunch of other countries have a big head start and lower cost. You've had people that have been able to focus on this stuff before it was commercially viable.
Q: What's the motivation behind the North Korean hiring scams?
McLeod: Most of these scams are about generating income for the state, though some are espionage aimed at placing operatives inside tech companies. Either way, companies end up with employees who aren't who they claim to be, pulling data out of systems they should never have accessed.
Q: How does Larridin identify AI trailblazers in a company?
Larrison: We basically segment your entire workforce. We tell you who those AI power users are. We tell you what good looks like — what are the people doing best that others can learn from? I love a customer that looks at their sales team, identifies the sellers with the best close rate, and then models that behavior.
Q: What does "good" AI usage look like in a company?
Larrison: Good looks like I'm using the AI to supplement what I do. It's upleveling my skill set and allowing me to focus on value-add rather than just doing rudimentary type work.
Q: Is AI a wave or a tsunami?
Fubini: As a society, we're just going to experience this. It's the same way that the internet changed how commerce was going to work. We're going to see AI change how every single job is essentially done. It's been incredibly disruptive for certain people. It's going to be empowering for others.
Q: Do you think humanoid robots will go mainstream?
Fubini: Do I really need a robot that cuts my grass and folds my laundry? I'm skeptical that we're going to have this wave of consumer robotics products. I do think we're going to see robots that are purpose-built for lots of different environments — hospitals, factories, even public environments. But I don't personally believe in the consumer ones. Eventually, we'll get the Jetsons-like experience. But I think that's probably a long way off.
