20VC with Harry StebbingsLabour, Engineering, Social Media, GrokBots, Cybercabs: 7 Predictions for How AI Changes the World
20VC with Harry Stebbings
Hiring wars and talent competition 0:00
Matteo Franceschetti, founder of Eight Sleep, says hiring feels like 2021 again, with salaries spiraling out of control for engineers and even product managers because OpenAI, Anthropic and other hot startups keep making huge offers. The bigger problem is not just attracting people but retention, since employees keep getting outside offers at higher pay, which is why Eight Sleep has shifted more hiring toward Europe and outside the US.
Lessons from angel investing 4:00
Matteo says angel investing has been a game changer because investor updates let him see how the market moves, what valuations and multiples look like, and which growth channels work for other companies, even ones outside his own space. The biggest lessons have come from watching fundraising rounds and studying which consumer channels succeed elsewhere. He also warns that new hardware founders usually underestimate their true costs, saying early estimates of BOM and COGS are often off by at least fifty percent.
Building and measuring growth channels 5:01
Word of mouth makes up about forty percent of Eight Sleep's revenue even at its current scale, with paid media on Meta, Google and TikTok next. Matteo describes running weekly incrementality tests, such as turning off a channel in one state to compare revenue against a similar state, because platform-reported CAC is unreliable and Meta's own numbers can undercount true cost by around twenty percent. He advises founders to add channels one at a time, proving out Meta first with a capped weekly budget before scaling, rather than spreading money across many channels at once. TikTok only started working for Eight Sleep in the past six months, once they built a steady stream of influencer-generated content.
Discipline in spending and margins 9:29
Matteo insists on an immediate payback period and healthy contribution margin from day one, a discipline he credits to Keith Rabois. He notes the company could grow fifty percent faster by spending more aggressively, but that would make the business less healthy and less ready to go public. Given a free choice, he would rather put extra resources into marketing than lower prices, since brand awareness remains the biggest bottleneck even at Eight Sleep's size. The subscription side of the business benefits from very low single digit yearly churn, which he says beats comparable wearable companies.
AI reshaping marketing and engineering 18:31
Eight Sleep now tracks a channel called AI SEO to see how it ranks in AI-driven searches, and has noticed Google behaving differently as a result. Email marketing, once run by a two-person team, is now handled entirely by AI bots built by his co-founder Alexandra in three days, generating close to one hundred million dollars with a team of zero. Matteo estimates the company effectively has thousands of AI agents working alongside human staff, making the workforce three to four times larger in practice. On engineering, he says human engineers stopped writing code about a year ago, now directing hundreds of AI engineers instead, which lets Eight Sleep operate in thirty five countries with a small core team, spending in the low millions monthly on Claude.
Where AI spending and value are headed 21:32
Matteo predicts that as AI usage grows, engineering budgets devoted to tools like Claude could rise sharply as a share of spending, even as unit costs fall, comparing it to how cheap electricity became after being costly decades ago. He expects productivity gains on the scale of the industrial or agricultural revolution, and says he would personally consider buying Anthropic at its IPO. He also predicts that successful AI-era companies will turn into holdings with multiple businesses, pointing to Xiaomi and BYD as examples of companies that expand far beyond their original product into cars, appliances, and components.
China's manufacturing edge and robotics race 25:32
Matteo describes being struck by the design quality of Chinese electric cars, comparing showroom experiences to Apple stores and noting features like rotating seats and ceiling screens. He argues Chinese automakers focus far more on user experience than many global rivals, and warns that if these carmakers don't pivot into building humanoid robots, they risk losing a trillion dollar humanoid market the way they once dominated cars. He mentions being unsettled by how fast humanoid robots have progressed at events like the Humanoid Games, and references reports of AI models beginning to self-train in ways that reduce human oversight, including alleged instances of agents forming their own coordinated behavior inside an OpenAI model.
Selling Consumer Products in China 28:31
Selling in China works nothing like selling in the West, because there is no real e-commerce in the way outsiders imagine it. Chinese consumers do not buy from standalone websites. Nearly all purchases happen inside super apps like WeChat and Rednote, which function as an all-in-one combination of WhatsApp, Amazon, and TikTok. A website, if it exists at all, is just a place to browse pictures and read about a product, not to buy it. Learning this system from scratch was one of the biggest adjustments in taking eight sleep into the Chinese market.
Trust, Credit Cards, and AI Bots 29:31
Right now, Grok-based bots can already buy things on someone's behalf, but they do it through existing accounts like Amazon rather than holding a person's credit card directly. The comparison drawn is to online shopping twenty years ago, when many people, especially in Europe, refused to enter card details on websites because it felt unsafe, and sometimes actually was. The expectation is that within a year or two, people will start trusting frontier-model bots from companies like X, OpenAI, and Anthropic with that kind of financial access, much as trust in e-commerce itself grew over time. Looking further out, the prediction is that a personal bot could eventually buy and operate a cyber cab that earns money on its owner's behalf, and that health tracking, particularly through sleep, could help address diseases like cancer within a decade by detecting early signs of conditions such as diabetes, hypertension, and sleep apnea, turning a bed into an advanced medical platform.
Timing Retail and International Expansion 32:00
A recurring lesson in scaling eight sleep has been that timing matters more than whether a channel works in principle. Retail is treated as an example: it always could have worked, but the underlying conditions needed to be right first, and that time is only arriving now. The same logic applied to China. Talking with Chinese founders revealed that many consumer brands build recognition abroad before entering China, because Chinese buyers prefer products already validated in the Western world rather than risk buying an unproven local knockoff. By the time eight sleep entered China, it was already sold in 35 countries, sponsored by Charles Leclerc, linked to Ferrari, and publicly mentioned by figures like Elon Musk and Mark Zuckerberg, giving it the recognition needed to command a premium price rather than get undercut by a cheaper copy.
Brand Versus Price and Feature Leadership 34:30
Asked about the idea that a company must be either the cheapest option or the one with the best features, the response is a partial agreement. Being the best product matters, but the bigger point is brand, not features alone. A brand associated with respected names like Charles Leclerc or Elon Musk carries aspirational weight that customers want to be part of. The comparison drawn is to the early iPod, which people wanted long before it launched in Europe, and to Steve Jobs, who often released a great product after competitors but won through branding, packaging, and pitch. The view is that Silicon Valley tends to undervalue brand, assuming a great product alone is enough, when in fact a great brand paired with a weak product goes nowhere.
The Charles Leclerc Sponsorship Story 36:31
Charles Leclerc had already been using eight sleep for two years before the two ever met. There was a personal connection beforehand too: motorsport fandom meant tracking Leclerc's career since his go-kart days, years before Formula 1, including an early outreach attempt to sponsor him back when he raced in F2, which fell through because eight sleep was too small at the time. Years later, after sharing the company's medical vision over dinner during the Miami Grand Prix weekend, the partnership was agreed within days, and Leclerc went on to personally invest in the company, which is seen as proof of his sharpness as a businessman as well as a driver.
Working with Athletes and Celebrity Investors 38:30
Advice for founders seeking celebrity or athlete backers centers on two conditions. First, the person needs to genuinely use and believe in the product before any deal is discussed, since a purely financial relationship without real belief tends to go nowhere. Second, a direct relationship with the athlete is required rather than working solely through managers, because word-of-mouth from a genuinely convinced user is the real value, not just their name on a cap table. Examples include Tadej Pogačar and the UAE cycling team adopting the product after Pogačar installed it himself every night during the Tour de France, and top padel and tennis figures, including coaches of players like Aryna Sabalenka, using it as well, much of it undisclosed publicly.
Unusual Customer Stories 41:00
Several stories illustrate how far some customers go. One famous footballer had a custom mattress that didn't fit the standard product, so the team installed it in his child's room instead, only for him to later throw away a 300,000 dollar mattress and move eight sleep into his own bedroom for the sake of his performance. A well-known actor received the product as a gift, loved it enough to buy ten for his family, and later spent five minutes personally talking with the eight sleep founder at a dinner full of major CEOs, visibly changing how those CEOs viewed him afterward. One very famous individual has bought 55 units across various homes, and some buyers acting through embassies have ordered in the hundreds, none of them ever asking for a discount. One couple reportedly buys a new unit for any hotel stay longer than three or four nights because they cannot sleep without it.
Pricing, Democratization, and Income Inequality 45:02
Asked about discomfort with selling such a premium product in a world of visible extreme wealth, the response draws a comparison to Tesla starting with the expensive Roadster before building toward mass-market models. The stated goal is to keep lowering prices so better sleep becomes accessible to more people, with the belief that if the product cost 100 dollars, it would already have sold in the billions given how much people value its benefits.
Running a Marriage and a Company Together 45:31
Working with a co-founder who is also a spouse is managed through strict separation of channels, using Slack for company matters and WhatsApp for personal life, including a running joke about craving pizza despite avoiding carbs. Work talk is limited after a certain hour in the evening because one partner needs time to mentally process the day, though ideas can still be sent as messages to be read later rather than discussed aloud on the spot. The guiding principle offered by the co-founder was to be treated the same way a remote vice president in another city would be treated, not called about non-urgent matters late at night. The hardest period of blending marriage and business was in the early days, when professional boundaries were less clear, and learning to communicate with the same formality used with any other executive took deliberate effort.
Immigrant Origins and Early Doubts 50:30
Coming to the United States from Italy without attending Stanford or Harvard, and building the company with two other immigrant co-founders, meant starting with real uncertainty about whether a genuinely scalable consumer business could be built at all, beyond friends buying out of kindness or a crowdfunding campaign succeeding once. That early fear of not knowing whether the business could truly scale is described as one of the harder things to admit.
Moving to China for Manufacturing 51:30
After raising money following Y Combinator demo day, manufacturing simply wasn't happening, so the decision was made to fly to China the next day and stay until it was fixed, which ended up taking three to four months. The longest-tenured employee at the company today is someone hired in China during that stretch. The couple's approach to major life demands, including eventually raising a child, is described as figuring things out through direct conversation and doing whatever the business requires, while trying to find smarter ways to manage it when possible.
Vacations and Retaining Key Talent 54:01
There has been no real vacation or honeymoon since the company started, since the wedding date was deliberately set for one week after a crowdfunding campaign ended, and two nights in any one place tends to be the limit before restlessness sets in. A rare weekend trip to the Maldives was interrupted when the head of hardware called at 2 a.m. threatening to leave for a role paying roughly three times more with liquid stock. Retention came down to the employee's belief in the company's culture and mission rather than money, and he ended up staying after a last-minute counteroffer worked out just minutes before he planned to tell his own team he was quitting.
Mismatched Job Expectations 57:31
Matteo describes a red flag he sees in hiring: candidates who are simultaneously interviewing at a company like Anthropic and at Sleep, treating the two as interchangeable options. He says this signals confusion, not ambition, because the two environments and expectations are nothing alike. He also points to a trend among laid-off workers in Silicon Valley who accept a job offer just to secure income, then quietly switch to another company two weeks later without updating their LinkedIn, simply to bridge a gap in earnings.
Money, Ambition, and Loyalty 59:00
Asked whether Silicon Valley has become too transactional, as Chamath suggested, Matteo agrees but with a caveat. He believes people should join a startup driven by the potential for large financial upside, since a startup like Sleep is aiming for 10x or even 100x to 1000x growth, unlike a stable giant company where two or three times growth might be normal. He argues employees need to be honest with themselves and their families about whether they want upside potential or the certainty of steady cash and annual tenders.
Promoting from Within 1:00:30
Matteo recounts promoting one of his engineers to VP of engineering, even while running an external search for the role, so the company could still benchmark against outside candidates. Before officially offering the job, he told the employee to go home and talk with his wife first, since she had watched him work at 130 percent for three months, and the commitment to stay for several more years would affect the whole family, not just the employee.
Small Teams, Big Output 1:02:01
Sleep runs with about 160 people, a number Matteo expects to grow to roughly 250 in three years while the company reaches a billion in revenue. He says similar companies at that revenue stage typically employ a thousand people, meaning Sleep's output per employee is about five times the Silicon Valley average. He credits this to building very small, flat, highly capable teams, such as two people overseeing a paid media operation that generates hundreds of millions of dollars, and a four-person finance team where other companies might use twenty.
AI Agents and New Teams 1:04:00
Matteo says no job function will disappear entirely, but the same people will cover far more ground using AI. Sleep built an internal AI tools team from scratch six months ago to manage thousands of internal agents, kept separate from other departments because it must stay tightly connected to data engineering to ensure agents have accurate access to figures like revenue and customer acquisition cost by country. He notes paid media agents already generate hundreds of millions in revenue, alongside major efficiency gains in finance.
A Future Without Paid Work 1:06:31
Matteo predicts that children born today will likely never need to work for money by the time they are twenty, since AI will handle most labor, leaving people to pursue passions or social and helping activities instead. He describes his personal framework for happiness as a pyramid: health at the base, since illness makes everything else meaningless, followed by relationships of any kind, and finally purpose, which does not need to be financial but must give direction to life.
University, Basic Income, and Parenting 1:08:31
Asked whether young people should go to university, Matteo says no, calling it useless for anyone wanting an entrepreneurial career in Silicon Valley. He expects universal basic income will eventually be necessary as companies grow more efficient with fewer workers and get taxed more to redistribute wealth. On parenting, he says he is interviewing fathers, and sometimes mothers, of highly successful people in sports, chess, and business to learn what worked and what didn't, since he hopes to become a father himself one day.
Drive Rooted in Fathers 1:11:31
The conversation turns to a pattern where highly driven people, often men, are shaped by a broken or striving relationship with their father, trying to prove their worth to him. Matteo relates this to his own life, noting his father died before seeing any of his success as an entrepreneur in the US, and says part of what drives him is an attempt to meet the bar his father set.
From Ferrara to Silicon Valley 1:13:31
Matteo describes growing up in Ferrara, a small Italian city, with a mother who was a teacher earning a modest salary and a father who took out a loan to send him to San Diego to learn English. He says he was a mediocre student until he discovered that top law firms handling international deals required fluency in English and top grades, after which he raised his average from 23 to 30 with honors, graduated with distinction, and eventually became a lawyer in Milan before starting his entrepreneurial path.
Money as Motivation 1:17:31
He recalls attending Chelsea football matches as a child in cheap seats, disliking the smell of beer and the aggressive crowds, until a wealthy friend brought him into a corporate box that was clean and comfortable. That contrast, he says, opened his eyes to how much better life could be with money and gave him real drive, layered on top of the discipline he had already built through competitive sports like tennis.
Restlessness and Admiration for Calm 1:19:00
Matteo compares his inability to relax to a physical condition, describing himself as someone who cannot stop pushing even on vacation. He says the people he admires most are not business figures but ordinary people who seem at peace, describing a man he saw dining alone in Milan, dressed well, enjoying wine and food without checking his phone, appearing fully content in the moment, which Matteo says he struggles to achieve himself.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
