I’ve spent my career at the intersection of three disciplines: operating executive, coach and dealmaker. That combination matters more than ever in frontier tech, where brilliant engineering teams often struggle with the last mile, turning breakthrough models into repeatable commercial motion, defensible pricing, and the kind of enterprise discipline that scales.
At Uber, I built partnerships and go-to-market across mobility, delivery, and new verticals (including Autonomous). Since then, I’ve worked alongside founders navigating the hardest part of growth: hiring the right people, defining what the company actually is, and figuring out how to sell something the world has never bought before. When I joined Empiric Earth’s board in April, I expected to advise from the outside. Six months later, I couldn’t stay there.
Here’s why.
The Trust Problem in Physical AI Is a “Real World” Problem
AI is leaving the screen. Autonomous vehicles, warehouse robots, and last-mile delivery bots are making real-time decisions in the physical world. They will only go as far as people trust them, and in the physical world, trust is earned on safety, not demos.
The hardest part of safety is the long tail: the near-miss at an intersection, the pedestrian stepping out between parked cars in the rain, the two seconds a driver looks away at exactly the wrong moment. These events are vanishingly rare in any single dataset, but they define whether a system is safe or dangerous.
Empiric Earth holds 10 billion miles of real-world driving data, 60 million safety-critical edge cases, and ingests 300+ million new miles every month from a live network of 350,000+ connected sensors across 98% of U.S. roads. You can’t simulate this. You can’t buy 11 years of history on the road. And you can’t shortcut the work of building a rights-cleared, provenance-tracked corpus where every scenario is matched to calibrated human baselines.
That corpus is the moat. But the business model is what makes it defensible.
Why Open Source + Governance = Competitive Advantage
I have learned from people like Peter Steinberger and the OpenClaw community that openness and security are not opposites; they’re complements. Peter showed me how radical transparency and developer-first design build trust faster than any sales deck. OpenClaw taught me that the best way to earn adoption is to give people control over their own data and let them see exactly how the system works.
An open-core playbook applied to physical AI helps harness the broader community to tackle critical problems together, with transparency.
Governance and Transparency Aren’t Optional. They’re the Product
Physical AI is different. When an LLM hallucinates, someone gets a bad answer. When an autonomous system fails, someone gets hurt.
That raises the bar for what a customer has to be able to trust. A model that performs well is only the start. A safety team, an insurer or a city also needs to know where the data came from, whether it was collected with the right permissions, and whether the company evaluating the system has a stake in the outcome.
Empiric Earth was built to answer those questions. We don’t build vehicles, and we compete with none of the systems we measure, so we have no reason to grade anyone on a curve. The data is de-identified. We also hold ourselves to a simple rule: if we can’t stand behind a number, it doesn’t ship.
That’s why governance and transparency have to be architected into the platform, not bolted on later.
Why More Women in AI Matters
I’m also here because I want to see more women building and leading in AI, especially as it moves into the physical world. The decisions we make now about safety thresholds, transparency standards, governance frameworks, and who gets to define “good enough” will shape this technology for decades. These frameworks will require an alignment to our values.
Empathy will play a huge role in that effort.
Those decisions are better when the people making them reflect the diversity of the people who will live with the consequences and align with their values. I want to help build the version of physical AI that starts with people and with keeping them safe. That’s the work here, every day.
Why Zach, Why Now
I’ve known Zach Greenberger since we were both building partnerships in mobility — me at Uber, him at Lyft. When someone you’ve trusted for years shows you a hard problem, an irreplaceable asset, and a once-in-a-decade market timing window, you pay attention.
Two companies, Nexar and Nauto, became one in September 2026. Mergers are hard. But when I look at what we’re building - a platform that can perceive, predict, prevent, and price risk across autonomy, commercial fleets, insurance underwriting, and municipal infrastructure - I see something that matters. Not just commercially, but morally.
A lot of people are afraid of what AI will do to their lives, and some of that fear is fair. I want to help build the version of this technology that earns trust through transparency, proves safety with evidence, and puts governance at the center, not the edges.
Commercializing Frontier Tech Is the Hardest Part
There’s a gap in frontier tech right now. Brilliant teams are building incredible models and systems, but many struggle to turn that into revenue, repeatable go-to-market, enterprise sales discipline, and the kind of commercial rigor that scales.
I’ve spent my career helping close that gap, turning partnerships into platforms, pilots into programs, and good ideas into businesses the world actually uses. That’s the work I do best: sitting with a leadership team and asking, “What needs to be true for this to work at scale? Who is the buyer? What is the workflow? How do we price this so it’s defensible and fair?” And then rolling up my sleeves and making it happen.
Empiric Earth has the data, the technology, and the independence. My job is to turn that into the commercial engine that gets this into the hands of every team that needs it and to do it faster than anyone expects.