Glenn Gow, The Scaling Executive Coach and host of The Scaling Executive Podcast, sits down with Jim Dukhovny, founder and CEO of Alef Aeronautics, to unpack why the world’s first flying car company draws a hard line between where AI belongs and where it doesn’t. When the cost of an engineering error is measured in human life, Dukhovny holds that AI must stay out of direct control of the outcome, because no complex real-world AI model today can be trained to meet the FAA’s one-in-a-billion fatal accident threshold, a standard that shapes every design decision at Alef Aeronautics.
This episode is for CEOs building products where an engineering error carries life-or-death consequences and who need a framework for deciding where AI belongs in that system.
Key Takeaways
- When failure carries life-or-death consequences, Jim Dukhovny holds that AI should not control the outcome directly today because no complex real-world model can currently be trained to meet the FAA’s one-in-a-billion fatal accident threshold.
- Alef Aeronautics uses AI heavily during aircraft design, including flow analysis and stress analysis, but routes every AI-assisted design through hardware testing before it reaches production, because the company does not extend AI trust from the design phase into the operational phase.
- Alef operates with four co-founders serving as officers and board members rather than a traditional management hierarchy, which forces major decisions to reach consensus among the founding partners before the company moves forward.
- Alef Aeronautics has secured over a billion dollars in pre-sales, a figure Dukhovny describes as delayed revenue built on a deposit-to-conversion model similar to other EV manufacturers, not cash already in hand.
Why Jim Dukhovny Keeps AI Out of Flight Controls
Jim Dukhovny told a keynote audience at one of Silicon Valley’s largest AI conferences that AI is bad, a statement he immediately narrowed to a specific context: controlling an aircraft. His reasoning rests on two constraints particular to Alef’s flying car. The FAA defines safety as one fatal accident in one billion, a bar Dukhovny says no team can currently train a complex real-world AI model to clear. Alef also faces a second constraint that self-driving car companies do not face at the same intensity: the price of an error in a flying vehicle is categorically higher than the price of an error in a ground vehicle.
“AI should not be used at least to control an aircraft today,” Dukhovny said, framing safety not as a top company priority but as a precondition beneath all priorities. “It’s not even number one,” he added, describing safety as the baseline the entire business depends on rather than one goal competing against others.
That does not mean AI stays out of the building entirely. Dukhovny draws a second line between design and operation, applying AI’s strength at complex inference directly to designing the flying car itself, including flow analysis and stress analysis during engineering. But he stops there. Every AI-assisted design must clear a testing layer before it reaches production. “We don’t trust AI completely, so we put every AI-assisted design through extensive hardware testing before it goes into the aircraft,” Dukhovny said, describing the gap between what AI produces and what Alef is willing to certify.
Why Alef Aeronautics Uses a Founder Partnership Instead of a Management Hierarchy
Alef operates with four co-founders, Dukhovny, Oleg, Constantine, and Paul, who all serve as officers and sit on the board of directors. Dukhovny describes the structure as a partnership rather than a management hierarchy, built around each founder’s specific area of expertise. “You don’t want to go somewhere where you’re not an expert,” he said. “You want to listen and you want to trust other people.”
Four decision-makers is an awkward number for reaching quick agreement, and Dukhovny said that awkwardness became the mechanism itself. “That kind of tells you that you cannot just say I want it or I’m expert in this or something like that,” he said. “You have to convince others and without it you don’t move forward.” The requirement applies only to major decisions. Small and medium calls stay with the founder who owns that area of expertise, while large, company-defining decisions require the founders to talk until they reach consensus. Dukhovny treats that friction as protective rather than costly: getting four experts to agree before a make-or-break call has, in his account, kept the company from moving forward on decisions that hadn’t been fully pressure-tested.
Why a Billion Dollars in Pre-Sales Is Not the Same as a Billion Dollars in Revenue
Alef Aeronautics has secured over a billion dollars in pre-sales for its flying car, a figure Dukhovny is careful to frame precisely. “That’s a delayed revenue, right? That’s not like we got a billion immediately,” he said. Alef positions itself as an automotive company and follows a deposit-to-conversion model standard to that industry: customers place deposits, and a known percentage of those deposits convert into completed sales over time.
Dukhovny separates the demand driving those pre-sales from the financial discipline required to fulfill them. He credits science fiction, not Alef’s marketing, with creating the demand for a flying car. “We’re capitalizing on the product of extremely high demand, and we were not creating that demand,” he said. The company’s job is to convert that existing appetite into a business with sound unit economics. That discipline shows up directly in how Alef models its own pipeline: rather than booking the billion dollars as validated revenue, the company runs the deposit book against both a bear case and a bull case conversion scenario, sizing the business on the conservative end rather than the headline figure.
The Founder’s Framework
| Principle | What it means in practice | Named evidence from this interview |
| Safety is the precondition, not a priority | Treat the regulatory safety threshold as the floor the entire business sits on, not one goal competing with growth or speed | Dukhovny keeps AI out of Alef’s flight controls, a constraint tied directly to the FAA’s one-in-a-billion fatal accident standard that no current AI model can be trained to meet |
| Trust AI to design, not to operate | Use AI’s strength at complex inference during engineering and design, then route every AI-assisted output through independent testing before it reaches production | Every AI-assisted design at Alef, including flow and stress analysis, must clear hardware testing before it is certified for the aircraft, keeping AI’s role bounded to design rather than flight |
| Force consensus on decisions that can break the company | Structure ownership so major calls require multiple founders to agree, using the friction of an even number of decision-makers as a discipline rather than a flaw | Alef’s four co-founder, four-board-member structure has kept company-defining decisions from advancing until all four founders reach consensus, using that friction to filter out under-tested calls |
| Separate demand you capitalized on from demand you created | When outside forces (culture, media, science fiction) create demand for a product, build the financial model to convert that demand conservatively rather than assuming it validates the business | Alef sizes its billion-dollar pre-sales book against modeled bear and bull case conversion scenarios rather than booking the full figure as validated revenue |
Quotes from This Episode
- “AI should not be used at least to control an aircraft today.” — Jim Dukhovny, Founder and CEO, Alef Aeronautics
- “It’s not even number one.” — Jim Dukhovny, Founder and CEO, Alef Aeronautics
- “You don’t want to go somewhere where you’re not an expert. You want to listen and you want to trust other people.” — Jim Dukhovny, Founder and CEO, Alef Aeronautics
- “Those big decisions are important, especially for early startup. Those are make and break.” — Jim Dukhovny, Founder and CEO, Alef Aeronautics
- “We want to make sure that the financial model is solid with both the bear case and the bull case scenarios.” — Jim Dukhovny, Founder and CEO, Alef Aeronautics
Frequently Asked Questions
Should a CEO let AI make decisions in a safety-critical product?
Jim Dukhovny’s answer is context-specific: AI should not control outcomes where the price of error is measured in human life, because current AI models cannot be trained to meet regulatory safety thresholds like the FAA’s one-in-a-billion fatal accident standard. AI is still valuable earlier in the process, particularly for complex design inference, as long as its output passes through independent testing before it reaches production.
How should a CEO structure decision-making among multiple co-founders?
Dukhovny’s approach at Alef Aeronautics gives each founder authority over their specific area of expertise for day-to-day decisions, while requiring all founders to reach consensus on large, company-defining decisions. He treats the friction of getting multiple people to agree as a safeguard against bad decisions rather than a bottleneck to route around.
How should a CEO interpret pre-sales figures when reporting company traction?
Dukhovny treats pre-sales as delayed revenue, not cash in hand, and models both bear and bull case conversion scenarios rather than assuming deposits will convert at a favorable rate. The distinction matters most when a company benefits from outside-driven demand, since capitalizing on that demand responsibly requires conservative financial modeling rather than assuming the demand itself validates the business.
Executives Work with Glenn Gow to Scale Their Companies and Careers
Glenn Gow is The Scaling Executive Coach — he coaches ambitious executives into the CEO seat and CEOs into successful exits. With 25 years as a CEO and 5 years in venture capital, Glenn helps leaders scale their companies by scaling themselves first. If this conversation was useful, you can apply for executive coaching with Glenn Gow or apply to be a guest on The Scaling Executive Podcast.
