Democracy Is the Platform Beneath Every Startup
Harrison Miller on Amazon’s coin-toss chaos, building through bubbles, and why democracy is the platform behind every startup
This week, we’re talking:
Building an AI startup is 95 percent company-building and only 5 percent AI 🛠️
How $60 million in rotting toys led to the deal that seeded Marketplace and AWS 🪰🧸
Religions have always been early adopters of technology, but the latest wave of faith-tech is built on sycophantic LLMs that risk turning “God” into a mirror 🙏🪞
China’s flagship AI model generates worse code for groups Beijing disfavors 🇨🇳🤥
Why ABC’s suspension of Jimmy Kimmel Live! is ripped straight from Viktor Orbán’s playbook 🇭🇺 🤫
My Take:
Building an AI startup is 95 percent company building and only 5 percent AI. If you don’t master the fundamentals—recruiting, culture, product market fit, go to market—you’re dead in the water, no matter how good your model is.
To separate hype from what endures, I sat down with Harrison Miller. He joined Amazon in 1998 when it was still “before dawn on Day One,” and in a coin toss wound up running the brand new toys business.
One Christmas he crushed it; the next January he was staring at $60 million in unsold inventory “rotting like fruit.” Later he helped engineer the deal that seeded Marketplace and laid the groundwork for AWS.
After Amazon, he became a growth investor. Today, he is working on a different kind of platform: democracy itself.
This conversation covers jungle gyms, bubbles, character when it counts, and why democracy is the platform every startup stands on.
TC: You joined Amazon in 1998. Set the scene—what did the company look like?
HM: Well, Amazon had just gone public, revenue was about $400 million (up from $150 million the year before), and we were growing super fast. But as Jeff would put it, it was still “before dawn on Day One.” For Jeff, “Day One” meant you should always run the company like a startup. Restless, urgent, never complacent. At that point Amazon was basically just books, with CDs and movies newly added. Two distribution centers handled pretty much everything but they were already bursting at the seams. I think most entrepreneurs would’ve paused to consolidate. But Jeff was already pushing to make Amazon “the everything store.” I came in as a general manager without a category. After that first holiday season, Jeff picked two new ones: electronics and toys. And in what felt like a coin toss, I got toys.
When I joined, the stock went from $20 to $120 in three months. We were hailed as geniuses. Jeff was Time’s Person of the Year. But Jeff kept telling us: you are not six times smarter than you were two months ago.
TC: So you went from books and CDs to toys?
HM: Exactly. We went from shipping CDs and paperbacks to shipping jungle gyms. Toys were a $20 billion industry so you can understand the appeal. But we learned how tricky they are fairly quickly. About 55% of sales happened in a five-week holiday window. You had to select and purchase product in February, then pay for it before it even went on the water from China, and if it didn’t sell, you were stuck. No returns. And because we were “Earth’s biggest store,” we obviously had to carry everything—from Barbies to backyard playsets. That had massive implications for cash flow and for operations. Meanwhile, we were still crammed into a ramshackle office in downtown Seattle, above a wig shop and a needle exchange.
TC: Ya, people always talk about Amazon’s frugality. Bezos made people use doors for desks as a cost savings measure and all that—
HM: The folklore was real, but the bigger point was clarity. Jeff hammered home a small set of principles: customer obsession, data-driven decision-making, bias for action, constant innovation, frugality, and what he called not optimizing for appearances. Basically, don’t waste energy on politics or looking good in the moment. Focus on substance over optics. You were interviewed against those principles. You were reviewed against them. If you performed really well but didn’t fit the culture, you didn’t last long. But the principles sometimes clashed with ambition. “Lean and mean” works when you’re shipping books. It gets messy when you’re suddenly storing and shipping jungle gyms and TVs. That contradiction was the crucible.
TC: It’s easy to forget with how successful Amazon has become, how precarious those early years were. What did the scary moments look like from inside?
HM: When I joined, the stock went from $20 to $120 in three months. We were hailed as geniuses. Jeff was Time’s Person of the Year. But Jeff kept telling us: you are not six times smarter than you were two months ago. This will reverse. And it did. Two years later the stock was at $6. Moms were calling my employees: “Quit. Go work at Boeing. This company won’t make it.” In toys, we nailed that first Christmas. But January hit. I was staring at $60 million worth of unsold inventory. And toys in January are like rotting fruit, devaluing fast. That forced us to run the first true unit-economics analysis…
TC: And the math didn’t work?
HM: The math said: we’ll never make money this way. And this is where the strong-form GM model kicks in. I owned the P&L and the launch date. It was on me to make the economics of the toy business work. I couldn’t just lay it at Jeff’s feet or blame “the everything store” ambition. If toys didn’t pencil, I had to change the model. So we’re looking across the street at Toys “R” Us. They were a much bigger company than us, but they were struggling to get a website off the ground. That’s when I pitched them: stop trying to build your own site, redirect your traffic to Amazon, and we’ll sell your toys on our platform. They said yes. Overnight we flipped from losing $60 million a year in toys to making $60 million.
TC: That sounds a lot like the beginnings of Marketplace, right?
HM: Close, but not quite. Marketplace came later. But this Toys “R” Us deal was the first time we made real money selling inventory we didn’t own. That deal opened the door to Platform Services—we ran Target.com end-to-end with Amazon’s stack: website, distribution, recommendations. Hugely profitable, but also a forcing function to rebuild our messy, fragile codebase into a services-oriented architecture. Andy Jassy later had the better idea: don’t sell the whole stack, sell it in pieces “by the drink.” That became AWS.
What helped was credibility: I’d been in the trenches, making numbers, sweating inventory, shipping products. And I knew the hero is the doer, not the investor. If you want the midnight call when something goes wrong, you have to earn it.
TC: So let’s fast forward a bit. You leave Amazon and move into investing. What lessons from those high-pressure years were useful as you changed course?
HM: At Summit Partners, we invested in founder-led, usually profitable companies. They didn’t need capital, so choosing us was a hiring decision as much as a financing one. What helped was credibility: I’d been in the trenches, making numbers, sweating inventory, shipping products. And I knew the hero is the doer, not the investor. If you want the midnight call when something goes wrong, you have to earn it. The hardest adjustment for me personally was the lack of scoreboard. As an operator, you know every quarter if you’re succeeding. As an investor… not so much.
TC: Right! And the maddening part is it can take five to ten years before you really know if you’re any good at it. I don’t have the temperament for that. The scoreboard being “TBD” for years would drive me nuts.
HM: Exactly. It takes a certain patience. A lot of investors act like they’re giants while the scoreboard still says TBD. Because the truth is, you don’t really know until years later.
TC: And yet here we are in an environment where nobody’s waiting years for anything. With AI, the money, the headlines, the valuations—it’s all happening in real time, almost like a speed-run of the dot-com bubble. You’ve lived through that kind of frenzy before. What’s the no-regrets advice you’d give to builders trying to keep their footing when the ground is shifting under them?
HM: Bubbles make building harder. Valuations are inflated, talent is crazy expensive, and hype can hide bad fundamentals. AI is intoxicating because it really does make all things new, but the same was true of the internet, and before that the steam engine. Every wave feels unprecedented, but the business basics don’t change. The fundamentals are still the fundamentals: real product-market fit, proper capitalization, disciplined decision-making. But the biggest lever is talent density. At Amazon, the mean level of talent was very high and the variance around the mean was very low. That mattered more than anything else. And talent isn’t just IQ or credentials. It’s character: courage, grit, the ability to stay mission-focused when things are going sideways. Who do you trust in the foxhole at 2:35 on a Tuesday afternoon when the system is down and customers are furious? That’s what separates A players from everyone else.
TC: It’s a strange state of affairs. Within a 60-mile radius of here, most conversations hinge on numbers and market sizing and carefully modeled capability. Those things matter, but we rarely obsess over character. I’ve met five-sigma entrepreneurs who are brilliant, off-the-charts talents. But sometimes we’re just not a match in terms of how they carry themselves. Not illegal, not unethical, just not how we get things done. And I’ll sound a little curmudgeonly here, but it worries me that in Silicon Valley the way we treat ourselves, our employees, our partners, that can get lost in the wash. There are lots of ways to get rich. But at what cost?
HM: I’ve seen the same thing. Being good people, aligned on winning the right way, however you define that, can be incredibly powerful. It doesn’t make success guaranteed, but it makes it sweeter and a hell of a lot more sustainable.
TC: And you’ve got to enjoy the ride too.
HM: Exactly. Life is too short to be surrounded by assholes.
Democracy is the platform under everything. In tech we talk about AWS, iOS, or air traffic control: systems you build on without thinking about them. They’re invisible until they wobble. And when they wobble, everything built on top wobbles too. Our entire industry, from innovation and wealth creation to the contracts you sign, rests on the assumption that democracy keeps working. That’s not a safe assumption anymore.
TC: Let’s switch gears. You’ve built companies, you’ve invested, but now you’re putting serious time and money into politics through Unite & Renew. It’s not glamorous work. I think a lot of people might be looking at your career and thinking: why even take that on?
HM: Unite & Renew pools long-term political capital from entrepreneurs and investors, folks like you, Tom, who’ve each made decade-long commitments. I run the fund as GP, and our mission is to make a concrete difference in what I see as the fight of my generation: leaving our kids a democracy that’s secure, functional, and trusted. This isn’t philanthropy. It’s politics.
TC: And that matters for people who are heads-down building companies because…?
HM: Because democracy is the platform under everything. In tech we talk about AWS, iOS, or air traffic control: systems you build on without thinking about them. They’re invisible until they wobble. And when they wobble, everything built on top wobbles too. Our entire industry, from innovation and wealth creation to the contracts you sign, rests on the assumption that democracy keeps working. That’s not a safe assumption anymore.
TC: So democracy isn’t just “nice to have.” It’s infrastructure, the platform beneath every other platform.
HM: Exactly. Think about the sweep of human history. For 100,000 years, every person on earth lived under some form of strongman. Kings, priests, warlords, whoever was “the guy.” Your survival depended on staying in his favor.
TC: So democracy is the anomaly, not the default.
HM: That’s right. Only about 250 years ago, basically three human lifetimes, America tried something radical: self-government by free and equal people. Imperfect, yes, and we’ve stumbled badly at times. But in just a few generations that experiment has produced more innovation, prosperity, and human flourishing than all prior systems combined. And right now, that platform is wobbling.
We’ve been backsliding since the early 2000s. Look at Hungary or Turkey. You get a populist strongman, the cousin of the prime minister suddenly has his hand in every contract, and pretty soon it doesn’t matter how good your product is. What matters is whether you’re in favor with “the guy.” That kills innovation and drags whole economies down.
For a startup it’s product-market fit, capitalization, and talent density. For a democracy it’s the rule of law, checks and balances, and elections we can trust. Strip away the fundamentals, and no amount of hype or capital can save you.
TC: So when people romanticize Hungary as a model, what they’re really praising is a medieval court economy, favor at the top instead of merit at the bottom.
HM: Exactly. It’s about favors flowing from the top, not performance. And the proof is in the numbers: Hungary went from a middling European economy to the bottom decile. That’s what happens when rules are replaced by the whims of one man.
TC: It reminds me of something I told some classmates at a reunion. Back in college, I was the geek saying, “Forget history and politics, study the math and computer science.” Now I realize the best idea I’ve ever heard isn’t a piece of code. It’s self-government by free and equal people.
HM: And it’s fragile. The founders warned about faction, polarization, as the greatest threat to the project. If citizens come to see each other as enemies, someone will always ride in and promise, “I alone can fix it.” That’s the road back to strongmen.
TC: So the danger isn’t just chaos. It’s that division creates an opening for the “savior” figure who promises shortcuts.
HM: Right. And if we’ve learned anything from history, it’s that shortcuts to power usually come at the expense of the rules that keep us free.
TC: Which brings us full circle. Whether you’re building a company or a country, the fundamentals matter more than the frenzy.
HM: Exactly. For a startup it’s product-market fit, capitalization, and talent density. For a democracy it’s the rule of law, checks and balances, and elections we can trust. Strip away the fundamentals, and no amount of hype or capital can save you.
TC: That’s the through-line here. From Amazon’s jungle gyms to AI to democracy itself—the basics are boring until they’re gone.
HM: And by then it’s too late. Which is why we have to fight for them now.
My Stack:
Finding God in the App Store 🙏🪞
Bible Chat has 30M+ downloads. Hallow outranked Netflix in the App Store. Some apps even let you “talk to God” directly. Others promise a priest, rabbi, or imam chatbot on call 24/7. The faith-tech boom is real and real profitable.
It’s a predictable story: religions have always been some of emerging technology’s earliest adopters. The printing press kept Judaism alive in diaspora with cheap siddurim and Talmuds. “Televangelism” made Christianity a broadcast empire. In the 2000s, Egypt and Kuwait lit up with SMS “fatwa hotlines” where people texted scholars for guidance.
I’m not against people seeking spiritual nourishment through whatever medium works. Trouble is, this latest wave of faith-tech runs on LLMs prone to sycophancy. What happens when “God” just +1s every thought you have?
Source: NYTimes
DeepSeek’s Double Standards 🇨🇳🤥
Tech has always carried political weight. The Soviets logged typewriter fonts. East Germany licensed photocopiers. Control the tools, control the story.
DeepSeek took it further. A CrowdStrike/Washington Post investigation found the Chinese AI model generated sloppier, less-secure code when the prompts came from groups Beijing disfavors. Less old-school censorship, more silent sabotage. Until the press dug in, nobody even knew it was happening. Yet another reminder why the fourth estate is so damned important right now.
Source: Washington Post
Free Speech Fallout 🇭🇺🤫
It’s troublingly ironic to mourn Charlie Kirk as a beacon of free speech and then use his violent murder to justify stripping others of theirs. ABC yanking Jimmy Kimmel Live! off the air harkened back to my conversation with Harrison about the Viktor Orbán playbook: Weaponize the levers of government for individual gain. Pressure media owners to fall in line. Punish those who resist. Reward those who comply. That’s how Viktor Orbán captured Hungary’s press. And, as former Hungarian MP Gábor Scheiring put it to CNN, “this story is very familiar.”