Oct 7, 2025 · 40m · y-combinator
Ask These Questions Before Starting An AI Startup · Y Combinator
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Jordan Fisher delivers a Y Combinator presentation exploring the strategic, technical, and ethical paradoxes facing AI founders in an era of rapid technological acceleration. He urges entrepreneurs to plan for AGI arrival within two to three years, build authentic trust through robust agent alignment, and focus on solving inherently complex problems with durable defensibility.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The partners as informed peer 0.0 Guest teaching 0.0 Guest disagreement 1.2 The partners pushing back 0.0
speaking balance: gold is the partners, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 37:15 Dismissing VC foresight as lagging groupthink
Fisher delivers his sharpest critique of the tech establishment, directly asserting that despite claiming to be forward-looking, VCs who think they are ahead in AI are already two years behind.
Hardest push from the partners ▶ 30:10 Audience member re-framing startup purposeAn audience member gently challenges the standard YC mantra by suggesting 'build something the world needs' is more critical than just 'build something people want.'
Biggest teaching moment ▶ 35:45 Deconstructing sycophancy in user preferencesFisher breaks down the fallacy of superficial user alignment, explaining how users pick sycophantic responses in isolation but reject sycophancy when presented with underlying operational principles.
The partners hold their own ▶ 31:50 De-bunking passion as the driver of startup resilienceFisher counters the conventional advice of pursuing startup passion, noting that after months of 100-hour work weeks founders will hate any idea unless driven by tangible impact and team commitment.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The partners as informed peer | Guest teaching | Guest disagreement | The partners pushing back | Why |
|---|---|---|---|---|---|---|
| The Core Paradox of Running an AI Startup | 0 | 0 | 1 | 0 | Jordan Fisher delivers a solo presentation exploring the paradoxes of startup focus and why founders must plan two to three years ahead for AGI. Because this is a monologue presentation without an active interviewer, host scores are strictly zero. | |
| Evolution of the Enterprise B2B Buy-Side | 0 | 0 | 2 | 0 | Fisher challenges the conventional wisdom that enterprise adoption will be slow, arguing the buy-side will leverage AGI agents to rapidly build or digest software in-house. As a monologue, host metrics remain at zero. | |
| On-Demand Code Generation and the Trust Factor | 0 | 0 | 1 | 0 | Fisher discusses the future of on-demand code generation and generative UI, highlighting the massive trust hurdle of letting LLMs modify database layers on the fly. Monologue format dictates zero host participation. | |
| AI-Native Team Structures and Organizational Dynamics | 0 | 0 | 1 | 0 | Fisher examines how AI-native teams will be structured and questions whether retrofitting existing products or building from scratch will win out across different verticals. No host interaction occurs. | |
| Trusting Automated Teams and Internal Guardrails | 0 | 0 | 2 | 0 | Fisher outlines the breakdown of traditional human guardrails when small, automated teams can make unchecked rogue decisions without whistleblowers. Host metrics are zero in this monologue. | |
| AI-Powered Auditing and Public Commitments | 0 | 0 | 1 | 0 | Fisher proposes self-deleting AI auditors as a potential mechanism for startups to verify compliance and mission alignment without leaking intellectual property. The monologue contains no host pushback. | |
| Alignment Requirements for Long-Horizon Agents | 0 | 0 | 1 | 0 | Fisher explains how economic demand for long-horizon agents running unmonitored for days or weeks will forcefully drive progress in alignment research. Host scores remain zero. | |
| Scaling AI Applications Under Compute Constraints | 0 | 0 | 1 | 0 | Fisher discusses technical moats during hardware capacity bottlenecks, citing specialized tacit knowledge like semiconductor fabrication at TSMC or ASML as defensible positions against LLMs. Monologue format. | |
| Intelligence Ceilings and Task Saturation | 0 | 0 | 1 | 0 | Fisher discusses task saturation ceilings and asks whether model neutrality mandates will be required as centralized frontier labs govern AI capabilities. Monologue format. | |
| Concluding Remarks: Changing the World vs. Short-Term Profit | 0 | 0 | 2 | 0 | Fisher expresses disappointment in founders focusing purely on short-term monetization rather than building enduring societal value during humanity's pivotal shift to AGI. Monologue format. | |
| Q&A: Information Sources and Curation Strategies | 0 | 0 | 0 | 0 | An audience member asks about information sources for building mental models on AGI, and Fisher advises rigorous curation of Twitter for intellectual exploration. The exchange is collaborative and respectful. | |
| Q&A: Post-AGI Economics, Universal Basic Compute, and Power Dynamics | 0 | 0 | 1 | 0 | Audience members ask about post-AGI value of money and individual alignment; Fisher highlights policy dilemmas around Universal Basic Compute and avoiding sycophantic model behavior. Interactions are purely collaborative Q&A. | |
| Q&A: Tech Groupthink and VC Blindspots | 0 | 0 | 3 | 0 | Fisher calls out severe groupthink in tech and VC culture, remarking that VCs claiming to be ahead of the curve in AI are actually two years behind. The audience member prompts this warmly. | |
| Q&A: Agent-to-Agent Communication Protocols and Game Theory | 0 | 0 | 0 | 0 | An audience member asks about inter-agent protocols like Google's HTA, and Fisher explains the complex implicit game theory required for seemingly simple tasks like calendar scheduling. The session concludes cordially. |
Statements from this episode (25)
Fisher: AGI is extremely likely to arrive within two to three years
“Cause it's extremely likely that we will have AGI in the next few years. Maybe it's not two, maybe it's three.”
Fisher: Enterprise buyers will adopt AGI agents within a few years
“And I think that's actually extremely nearsighted because what's going to happen, I think open question is the buy side, the enterprises, they're going to get armed with AGI or strong agents over the next couple of years as well.”
Fisher: Large enterprises are replacing SaaS purchases with AI-assisted internal builds
“Large enterprises sometimes aren't even going out and buying software from SAS providers anymore. They're like, I can just like Throw two people at cloud code and they'll build it and it'll be dedicated to the capabilities that I need for my organization.”
Fisher: AI code automation may raise software quality bar significantly
“I think another outcome actually is the opposite. It's like maybe all of this automation on, on generating code makes it easier to just raise the quality bar extremely high.”
Fisher: Current AI Is Not Trustable Enough for On-Demand Backend Code
“And obviously right now AIs are not trustable enough to make that happen.”
Fisher: Retrofitting Existing Products With Distribution May Beat AI-Native Startups
“It's natural to think it's better to build it a new product from the ground up that's AI native. Like that's like my startup mentality, like new technology revolution. You need to start from scratch if you want to build this right. But that might not be true. …”
Fisher: The definition of an AI-native company changes every 6 to 18 months
“And actually that might be different every six, 12 or 18 months, right? Cause the capabilities of AI are changing. So like an AI native company today might be different than what an AI native company tomorrow looks like in 12 months. You might be outdated if y…”
Fisher: Users want a single AI assistant bridging personal and professional work
“As a user, you want one agent for all of your things, right? So me, I want my personal agent and my professional agent at work to be able to work together, right?”
Base model alignment does not guarantee AI agents align with user interests
“The truth is, even if you have a perfectly aligned, you know, quote unquote, like intent aligned model, It's going to be used by a corporation or a startup to build an agent for that user. And then the question is, can you trust the startup that's building the…”
Fisher: Traditional corporate trust relies on employee whistleblowers checking bad leadership
“One of the core reasons we trust companies today is because they're composed of a diversity of people. You can trust to some extent that if there's reasonable culture at a company, that if the company decides, if the CEO decides to do something bad, that someo…”
Semi-automated companies allow single bad actors to act without human whistleblowers
“In a semi-automated world, that's no longer true. And it could be the fact that a single person could make a decision that changes the entire impact of a product. And there's no single person that might be aware of that except themselves. It gets extremely eas…”
AI auditors hold advantages over humans because they can delete their memory
“There's this huge advantage that AI has over humans when it comes to bringing comfort to an audit, which is that they can be less biased and they can also have no memory.”
Fisher: Continuous AI auditing of company operations will soon be possible
“Maybe this isn't the right way to do it, but I think things like this are going to be possible soon. They're not possible today, and they might be the flavor of things that we need to start building trust once we're in this world where we've lost a lot of trus…”
Fisher: Economic pressure for long-horizon agents will drive AI alignment progress
“I'm actually really positive and bullish that there is this economic pressure in a good way To make progress on alignment because long horizon agents require it.”
Fisher: Many AI developers have abandoned fine-tuning for context management
“I think a lot of people have abandoned fine tuning and said, actually, I'm just going to do better context management.”
Fisher: Technical compute optimization offers startups a competitive advantage for 1-2 years
“So I think this is like a place where if you're interested in the technical details, you can have a competitive advantage, at least for the next year or two because capacity is really going to matter, right? You know, often from a product perspective, I like t…”
Fisher: Hard physical and semiconductor problems provide durable moats post-AGI
“And I think a lot of things actually, like TSMC, like ASML, those are hard problems that eventually will get easy with robotics, et cetera, but robotics are lagging behind. So I think there's these hard problems that will exist even in two years, and if you're…”
Fisher: Tasks with intelligence ceilings face rapid commoditization pressure
“If there is a ceiling, then the commoditization for that task is going to hit much sooner, right? You actually can't stay on the edge longer by moving to the next model, figuring out how to prompt the next model. The task saturates, right? So the commoditizati…”
Fisher: AI Model Providers Will Arbitrate What Gets Built
“If we end up relying on these models, that's a massive question for society, right? That there's going to be a handful of corporations that get to decide what is okay and not okay for an AI to do for you. And if we start relying on AI to do everything, then th…”
Fisher: Lasting AI startups require defensibility, though quick flips do not
“I honestly think there's a lot of money to be made in the next, in On a six to 18 month horizon, if all you're optimizing for is, like, you know, hockey stick curve, grow your ARR, flip the company and make a quick buck, like you don't, you might not need some…”
Fisher: Post-AGI capital won't need labor buy-in, worsening wealth concentration
“But once AGI arrives, you don't need labor buy-in anymore. Right. Like you don't need folks like me to approve of the thing that you're building or of your morals or whatever. Right. Like capital begets capital in that world. And that can easily spiral out of …”
Fisher: Users choose honest AI over sycophancy when asked at principle level
“But I think if you take a step back, and you say, hey, hold on a second, here's two principles, and you can choose the AI that follows this principle or this principle, The first principle is that we're never gonna blow smoke up your ass. We're only gonna tell…”
Fisher: VCs Investing in AI Are Already Two Years Behind
“And I think even today, like you talked to a bunch of VCs and they're like, oh yeah, like I'm ahead of the curve. Like I'm investing in AI. I'm like, no, like you're two years behind already. Right. Like, tell me about what you think needs to happen in two yea…”
Fisher: Blockchain ideas could address AI trust and compute distribution
“But I do think You know, in this world where we need trust, yeah, like, those, that's the right set of ideas. We need ideas like that in others, right? Like, you know, something that is always touching on is like, you know, AI-powered audits. Can, like, two di…”
Fisher: AI Scheduling Assistants Must Master Implicit Game Theory and Power Dynamics
“But actually, like, you've already lost the game of being a good personal assistant, because there's a game theory component to how you schedule meetings, right? And if you're, like, too liberal with, like, showing that there's free slots, you're, like, commun…”