Summary
The episode "Startups You Can Now Build With AI" by Y Combinator explores the transformative impact of advancements in large language models (LLMs) and AI technologies on startup opportunities. The panel highlights the increasing feasibility of previously difficult or impossible startup ideas due to breakthroughs such as Gemini 2.5 Pro, boasting unprecedented features like million-token context windows. A major theme is the emerging demand for AI infrastructure and tooling, which remains underdeveloped despite the advances in model capabilities. Recruiting startups serve as a prime example where AI dramatically lowers barriers by automating candidate evaluation without massive labeled data, contrasting with historical attempts like TripleByte that required extensive manual effort. The episode also discusses how AI is simplifying complex multi-sided marketplaces into more manageable two or three-sided models by automating intermediary roles, thereby reshaping platform economics. In education, personalized AI tutors are viewed as a 'holy grail,' enabled now by LLMs to deliver highly tailored learning experiences, demonstrated by startups like Revision Dojo and Adexia. The panel notes the rapid decrease in the cost of intelligence, signaling a shift towards freemium business models and widespread AI adoption, analogous to how mobile phones revolutionized consumer markets. They also caution about the organizational challenges within large companies like Google, where fragmented AI initiatives create internal competition and user confusion. Google’s advantage in hardware, specifically Tensor Processing Units (TPUs), is highlighted as a key enabler for cost-efficient deployment of large context window models. On the consumer side, examples such as Meta's AI assistants illustrate difficulties in balancing innovation with user experience and privacy. A crucial innovation discussed is empowering users by making the system prompt editable, offering greater control over AI behavior. Reflecting on past waves of tech-enabled services, the podcast shows how AI agents might revive full-stack startup models by reducing operational complexity and improving margins. Overall, the episode emphasizes a paradigm shift in how startups should approach idea formation, urging founders to actively explore AI's new capabilities as the 'idea maze' has shifted significantly. The conversation is both a call to action and a practical examination of the opportunities, challenges, and evolving dynamics shaping the AI startup ecosystem today.
Key Takeaways
- 1There is a compelling and underexplored opportunity for startups to build robust infrastructure and tooling that facilitate scalable, practical AI deployment, especially around autonomous AI agents. Despite rapid advances in LLMs such as Gemini 2.5 Pro, the ecosystem lacks mature platforms that enable easy AI integration, monitoring, and operationalization.
- 2AI has transformed the recruiting startup landscape by automating candidate evaluation through LLMs and code generation models, eliminating the years-long data labelling and human interview bottlenecks that challenged earlier companies.
- 3AI-driven simplification of traditionally complex multi-sided marketplaces into leaner two- or three-sided models is reshaping platform dynamics by automating intermediary functions formerly requiring humans.
- 4Personalized AI tutors and hyper-personalization in education are becoming viable at scale due to advancements in LLMs, representing a transformative chance to improve learning outcomes and engagement.
- 5The dramatic reduction in the cost of computational intelligence through techniques such as distillation democratizes AI access and is likely to reinvigorate freemium and massive consumer adoption models.
- 6Startups need to recognize that the traditional 'idea maze' has been fundamentally reshaped by AI breakthroughs, requiring proactive exploration and experimentation rather than relying solely on historical startup heuristics.
- 7Startups that effectively replace entire enterprise teams with AI tools, rather than merely augmenting workflows, can command significantly larger budgets, accelerating revenue growth and setting benchmarks for future consumer AI monetization.
- 8Consumer AI startups, particularly in personalized education, face the challenging dynamic where the user differs from the purchaser, complicating monetization and requiring AI quality comparable to human tutors to gain paid adoption.
- 9Building durable competitive moats for AI startups requires more than AI model integration; it demands strategic brand development, high switching costs, and deep integration with existing ecosystems and workflows.
- 10Large tech companies like Google face internal fragmentation and cultural clashes between competing AI teams, leading to inconsistent product offerings and external developer confusion.
Notable Quotes
"“Every other week, we're certainly realizing there's a new capability, a million token context window in Gemini 2.5 Pro. It's just really insane right now. And the thing to take away from that, though, is that we have an incredible number of new startup ideas, some of which are actually very old, and they can only happen right now.”"
"“And one idea that is very personal to me would be recruiting startups since I ran a recruiting startup, TripleByte, for almost five years. We started with the thesis of you don’t just want to let anyone on your marketplace. You want to build a really curated marketplace that evaluates engineers and gives people lots of data about who the best engineers are. And this was all pre-LLM.”"
"“TripleByte raised something like $50 million. Our main competitor raised over $100 million. I think in aggregate hundreds of millions of dollars went into funding recruiting marketplace companies. And overall as a category did not do particularly well.”"
"“And their whole premise is to build AI agents that run the screening for technical interviews. Where a lot of engineers spend a lot of time just doing a bunch of interviews. And the pass rate is so tiny. When I used to run engineering teams at Niantic, all that pre-screening was just so much work. The engineers hate doing it.”"
"“You know, that's the moment where you could have 100 million or a billion people using it. OpenAI might be furthest ahead with it, but the hope is that, you know, really thousands of apps like this start coming out across all the different things you'll need. And that's something that I know we'll keep saying it. Like, it's going to happen.”"
"And when GPT-3 and 3.5, the early adopters of it, started coming out, they saw that, wow, this is going to be the moment. They doubled down and they've been on this trajectory now with lots of MAUs, EAUs that's really working out."
"Sometimes it's integration with other technologies that are sort of surrounding that experience. Like in a school, it would probably be being connected to Clever, for instance, like login is authentication is pretty obvious."
"I mean, there's definitely the sense that if two orgs are competing and fighting, normally in a normal org, you go up and in a functioning startup, for instance, you know, it goes up to some level. And then ultimately the CEO or founders, and then they just say, okay, well, I see the points over here. We’re going this way."
"It's like if Google replaced Google.com with Gemini Pro, it would instantly presumably be like the number one chatbot LLM service in the world. But it would give up 80% of its revenue."
"You just, you now you have an AI assistant that's just in all of your chats and you sort of, it comes with a, you can just at it and they will just start talking in a group chat. And it feels quite invasive actually. Well, it's not that smart. And then it can't do anything."
"And if you are actually going to empower your users, you actually allow your user to change the system prompt, which is the part that normally is like above, you know, to use the Venkatesh Rao's idea of like sort of the API line. And it's sort of like the system prompt is actually what is exerted, like sort of imposed upon the user. And so, you know, Gemini follows this very specific thing."
"That was like tech enabled services for recruiting. We also had Atrium, which was tech enabled services for law firms. It started with Balaji's blog post about full stack startups, if you remember. Like the concept was just that software eats the world means software just kind of goes into the real world. And so this is not the success example, but an example of it was, hey, like instead of just having an app to deliver food, you should also like have a kitchen that cooks the food and software to optimize the kitchen. And you just do everything."
"Basically, like the margins didn't work out particularly well. And so they need to keep raising more capital. And so if you were like a fearsomely good fundraiser, you could sort of do it and kind of push yourself. But even in those cases, I think most of those businesses at some point, it just caught up with them. Like at some point, like actually we have to figure out a way to scale the business and have good margins and make this like profitable and not just rely on the next fundraising round is what I felt hurt a lot of the..."
"If you look at within YC, we have Lagora, which is like this, like one of the fastest growing companies we've ever funded. And it's not building a law firm, but they're essentially, you know, building AI tools for lawyers. But you can see where that's going to extend out to you, right? Like eventually the agents are just going to do all of the legal work and they'll be the biggest law firm on the planet. And yeah, I think that's a kind of full-stack startup that just wasn't possible pre-LLM."
"Clearly, if you were working on ML ops in 2020 and you just stuck it out for a few years, you're in the right spot. You actually have a team that stuck it out. But I mean, part of the lesson is sometimes it will take a bit of time for technology to catch up. And this company called Replicate that you worked with stuck it out. It was from that era. Replicate was from winter 20 and they started the company right before COVID. And during the pandemic, it was going so poorly that they actually stopped working on it for several months and just like didn't work on it because like it wasn't clear that the thing like had a future at all. And then they picked it back up and just started like working on it quietly. But it basically was just like they were just building this thing in obscurity for two years until the image diffusion models came out. And then it just like exploded like overnight."