The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 83 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 1 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Tarpit ideas seem deceptively easy but conceal hidden structural barriers
“They all form around some, like, widespread problem that lots of potential founders encounter. And it's a problem that seems like it could be easily solved with a startup, but it's an illusion. There's actually a structural reason why it's very hard or impossi…”
Founder-market fit is the single most important criterion for evaluating startups
“If I had to pick like one most important criteria, it'd probably be this one. And what I mean by founder market fit is just, are you the right team to be working on this idea?”
The best startup problems are those where the customer alternative is nothing
“Like if you're, if the alternative to your solution is literally nothing, that's what a good problem looks like.”
Fintech and vertical SaaS have drastically higher unicorn hit rates than consumer
“Over the last 10 years, if you started a company that did, like, fintech infrastructure or vertical SaaS for enterprise, the probability that your company became a billion-dollar company was astonishingly high. Whereas if you started something in consumer hard…”
Boring startup ideas have a much higher success rate than fun ideas
“And the thing is that because most founders shy away from ideas like this, boring ideas like payroll software have a much higher hit rate than fun ideas like apps to find new restaurants to eat at, Or, like, apps to find the next song to listen to. Something l…”
Friedman: Hard-tech companies face high capital needs and feasibility uncertainty
“So the way I see it is a hard tech company is a company that fulfills two criteria. One, it will take a lot of time and money to build your first product. And two, even if you had lots of time and money, it's not clear if it would be possible to build it at al…”
Friedman: Software startups face market risk; hard tech faces technical feasibility risk
“So if your company is building a normal website or a mobile app, you probably have mostly market risk, which is to say you have a new idea. It's not totally clear if people are going to want this thing that you're making, but you probably don't have much techn…”
YC is the world's largest bio and hard-tech seed investor
“YC is actually the largest bio seed investor in the world and the largest hard tech seed investor in the world. This includes accelerators, seed funds, every kind of investor.”
Astranis built its first test telecommunications satellite for under $50,000
“It turns out actually that the cheapest telecommunication satellite that is useful costs like at least ten million dollars to build and launch. And so Astronus' hack was to start with a test satellite. The satellite in this photo was their first satellite. The…”
Friedman: Hard-tech startups can raise capital by preselling before building
“So basically they sold it before they made it, and this is a very generalizable technique that a lot of hard tech companies use in one form or another.”
Non-binding LOIs effectively filter casual interest from serious customer demand
“Because it's not binding, it doesn't actually commit the customer to buying, but because it looks like a contract, customers take it really seriously. It's easy when you're talking to a customer for them to, like, be polite and casually say, like, sure, I'd bu…”
Friedman: Top hard-tech founders minimize step sizes between funding rounds
“Really good hard tech founders are maniacal about, like, pushing down the size of each of those steps so that each step is as small as possible, which makes it as easy as possible for them to achieve the milestones that they need to raise the next round of fun…”
Friedman: Startup compensation has risen significantly, narrowing big-tech pay gap
“Compensation at startups has gone up a lot as well. And so I think we may be closer than they think, but there's still more that startups have to do on that front.”
Friedman: Ploy generates substantive content upgrades rather than AI slop
“Like, it's not just prettier, which obviously it is. Like, the, obviously, like, the visual design is much better, but actually, the content, It's better too. And we always hear about like AI slop, but like if it was AI slop, that wouldn't have worked.”
Friedman: YC Has Primarily Funded AI Companies Since ChatGPT
“For the last few years since ChatGPT, YC has been funding mainly AI companies”
Friedman: AI Agents Dramatically Increased the Volume and Complexity of YC Queries
“It didn't just make it easier to answer questions. It dramatically increased the number of questions that we would ask and dramatically increased the scale and complexity of the questions that we would dare to ask.”
Friedman: Variance emerges from stealth with $21M Series A
“We are announcing that Variance is coming out of stealth and announcing their twenty-one million dollar Series A, and I am really excited to be here with Karine, the co-founder of Variance.”
Friedman: AgentMail exploded in popularity after OpenClaw grew
“And so Agent Mail went the opposite way, and they built, like, the first email provider that's designed for AI agents, and it was doing well even before Open Claw, but once Open Claw got big, it was just, like, exploded.”
Friedman: Parahelp powers customer support for Perplexity, Replit, and Bolt
“They're actually powering the customer support for Perplexity and Replit and Bolt and a bunch of other like top AI companies now.”
Friedman: Meta-Prompting Is a Consistent Theme Among YC AI Startups
“Meta prompting, which is one of the things we want to talk about, because that's a consistent theme that keeps coming up when we talk to our AI startups.”
Friedman: YC uses a debug info field in agent response schemas
“In the response format. To give it the ability to have part of the response be essentially a complaint to you, the developer that like you have given it confusing or underspecified information and it doesn't know what to do. And then the nice thing about that …”
Friedman: Gemini API added thinking traces for debugging prompts
“Cause if you're just using Gemini via the API until recently, you did not get the thinking traces and like the thinking traces are like the critical debugging information to like, understand like What's wrong with your prompt? They just added it to the API. So…”
Solugen ships tanker trucks of hydrogen peroxide daily from Houston
“Today, they actually have this huge plant in Houston that ships out like tanker truck fulls of hydrogen peroxide like every day.”
Friedman: Close to 50% of YC Summer 23 Batch Used LLMs
“I think for summer of 23, it was close to 50% of the batch.”
Before Brex, banks strictly refused to issue corporate credit cards to startups
“And before Brax, if a startup NYC wanted a corporate credit card, they literally could not get one because no bank would give a credit card to a startup.”
Replit originally started as an interesting side project, not a business
“This is exactly how Replik got started. It was just something that Amjad found interesting. It wasn't supposed to be a startup originally.”
Many YC applicants pitch ideas completely unrelated to their actual domain expertise
“It is weird how many startups apply to YC and we look at their applications and the founders like are actually legitimate experts in something, but the idea that they're applying with is like something completely different.”
Jetpack Aviation pre-sold flying motorcycles online to prove customer demand
“A good example of this is a company called Jetpack Aviation from two batches ago, which is building the flying motorcycle in the picture. And what Jetpack Aviation did was they ran a pre-sale campaign, and they basically sold flying motorcycles to a bunch of p…”
It is illegal to conduct pre-sales for FDA-regulated medical products
“For example, if you're doing something medical that requires FDA approval, it's actually illegal to do pre-sales, so don't do that.”
Kyle Vogt built the original Cruise self-driving MVP in three months
“The original Cruise car Was built in less than three months during YC. Kyle basically just, like, was in a garage building this car and writing code for, like, three months solid, and by the end of YC he had an MVP that he could use to drive on the highway to …”
Friedman: Scribd hired half its early engineering team via Hacker News
“I think probably half the engineering team that we hired came through Hacker News, and that's not atypical for YC companies.”
Friedman: Scribd Was Originally Hosted on Dorm Room Servers Before AWS
“This is before AWS. This is on a physical server. Originally physically in my dorm room closet, which is how I learned that servers are loud.”
Dropbox had approximately 20 cloud storage competitors when it initially launched
“So when Dropbox launched, there were already about 20 cloud-based file storage companies. Dropbox was like the 20th company in its space to launch.”