Having a real-time video conversation with an AI is a trippy feeling. Talk about a "feel the AGI moment", it really does feel like the computer has come alive.
I think there is some risk of that. But commitment isn't static. A lot of incredible companies started with founders who were just toying with an idea and weren't committed at all, then they became more committed over time as things started to work.
We might, and I'm sure you're right that there are many great founders not applying to do YC because they don't want to move here.
But I think it would be a better analogy to compare YC to a university, rather than to a company. It's true that many companies operate remotely very effectively. But essentially zero universities have stayed remote since the early days of the pandemic.
Y Combinator (yes, the people who run this site) | Full stack web and AI | San Francisco | Onsite | Fulltime
YC itself hires software engineers from time to time. Many people don't know that YC has a small but very important software team. We run several websites - not just this one, but also workatastartup.com (the main way YC companies hire), a social network for YC founders, and the core infrastructure that YC itself runs on.
Recently, we've been spending most of our time building our own AI agents. Over the last year, YC has mostly funded AI agent companies, so it's been cool to go really deep building our own AI agents too. We're building agents to automate every aspect of what we do, and pushing the limits of what's possible with the current models to do it.
An unusual aspect of being on the YC software team is that you'll get full access to the YC program, founders and partners. The YC batch runs in the same building we work out of, so there are talks with famous founders happening most nights, right downstairs in our building. If you want to start a startup someday, working at YC would be an excellent jumping off point - several YC companies have come from team members who decided to start their own company.
YC works fully in-person in SF. We're happy to help you move here if you're in the US already. Unfortunately we can't sponsor new visas, but we're happy to transfer existing ones.
YC offers highly competitive compensation and benefits.
One of the interesting potential futures is one in which every software engineer can build awesome robotics systems.
What really unlocked ML was when you didn't need to have a deep academic background in ML to be able to use it. Now that the primitives exist, every software engineer can build a compelling ML app.
I'm not sure if robotics is there yet, but if not it probably will be.
That's true. But I don't think revenue multiple is the right metric to look at.
If you just look in terms of aggregate market cap, defense companies are some of the larger companies in the world - i.e., Lockheed Martin is valued over $100B. That's a good sign that it's possible to build a big company in the space, which is all you need.
A lot of the comments in here are about the debate on whether companies should be remote or in-person.
That's an important debate but orthogonal to what I wrote. The YC batch is not a company. It doesn't really have a close analogue, but if you forced me to choose, I'd say that doing YC is more similar to going to college than working at a company. And as all we all know, while many companies are staying fully remote, hardly any university is.
Having now done this back-to-back, I can tell you exactly the ways in which in-person YC turned out to be better than remote YC.
1). Most founders in remote YC didn't make strong connections with their batchmates. When I ask founders from remote batches "how many founders in your batch are you still close with?", they typically give an answer that's 0-3. When I ask founders from in-person batches the same question, it's 10+.
2). When YC really works, it's because it not only conveys some factual advice, but changes the way founders think and behave.
When founders go through in-person batches, they're usually significantly different by the end of the batch - tougher, savvier, and more formidable. Whatever causes that did not translate well to zoom.
3). In-person YC is simply more fun. YC has always been in part about being fun experience, because startups need to be fun or they'd be too difficult and demoralizing. Zoom is very effective for communicating information, but no one has fun at Zoom parties.
That's exactly right. When people see YC funding a lot of AI startups, a lot of them think it must be because YC has some thesis about AI.
Actually, it says something much deeper about the world than whatever YC's partners' opinions are. YC funds founders, not ideas, so the reason that so many companies in S23 are AI startups is that that's what founders want to work on right now. It's an emergent phenomenon, like stock prices in the market.
One thing that's interesting is that there have been many hype cycles between 2006 and now (chatbots, several waves of crypto, VR, online-to-offline, etc). YC funded a few companies in each of those hype cycles but never anything like the current %.
Perhaps other people with deeper AI knowledge can weigh in here too. But at the time, there were the two things that tipped me off.
1) Cyc's reasoning fundamentally did not feel "human". Cyc was created on the premise that you could build AGI on top of formal logic inference. But after seeing how Cyc performed on real-world problems, I became convinced that formal logic is a poor model for human thought.
The biggest tell is that formal logic systems are very brittle. If there is any fact that is even slightly off, the reasoning chain fails and the system can't do anything. Humans aren't like that; when their information is slightly off, their performance degrades gracefully.
2). Imagine a graph where time/money was on the x-axis, and Cyc's performance was on the y-axis. You could roughly plot this using benchmarks like SAT scores. It was clear if you extrapolated this that Cyc was never going to hit human-level performance; the curve was going to asymptotically approach something well below human-level performance.
As a side note, if you look at the performance of LLMs, I would argue that you get the opposite result for both criteria.
I was certainly interested in working at Cycorp full-time. But after two summers there, I could tell that the technical approach they were taking was just not working.
My first summer, I was an ontologist, which was a unique role that only existed at Cycorp where they hired people to literally hand-enter facts like "A cat has four legs" into Cyc using formal logic. My second summer I programmed (poorly) in Lisp for them.
Having a real-time video conversation with an AI is a trippy feeling. Talk about a "feel the AGI moment", it really does feel like the computer has come alive.