Jul 10, 2025 · 43m · y-combinator
Andrew Ng: Building Faster with AI · 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
At the Y Combinator AI Startup School, Dr. Andrew Ng delivers a keynote address on how artificial intelligence enables founders to execute at unprecedented speed, breakdown application opportunities, leverage agentic workflows, and build rapid feedback loops. Through his talk and Q&A session, Ng offers practical product development guidance while pushing back against exaggerated AI existential risk narratives and championing open-source technology.
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 1.5 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 ▶ 28:10 Rejecting AI extinction and compute hype
Ng forcefully calls narratives about AI leading to accidental human extinction ridiculous and calls out companies using fear for fundraising.
Hardest push from the partners ▶ 34:28 Audience researcher questioning brick accumulationAn audience researcher challenges Ng's static brick metaphor by bringing up dynamic token costs and time overhead.
Biggest teaching moment ▶ 30:25 Electric motor analogy for AI safetyNg educates the audience by explaining that safety is an application-layer property rather than an intrinsic technology-layer characteristic.
The partners hold their own ▶ 34:28 Technical research grounding in audience questionThe student researcher demonstrates deep subject domain grasp when contrasting static engineering accumulation with intent distribution dynamics.
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 |
|---|---|---|---|---|---|---|
| Execution Speed as the Key Predictor for Startup Success | 0 | 0 | 0 | 0 | Andrew Ng opens his keynote by framing execution speed as the primary predictor of startup success and mapping out the AI stack. | |
| The Rise of Agentic AI and the Orchestration Layer | 0 | 0 | 1 | 0 | Ng explains the technical significance of agentic AI and slightly critiques marketers for slapping the agent sticker on everything. | |
| Working on Concrete Ideas to Increase Speed | 0 | 0 | 0 | 0 | Ng discusses how concrete product ideas buy execution speed compared to deceptive, vague ideas that receive false praise. | |
| Rapid Prototyping and the Build/Feedback Loop | 0 | 0 | 0 | 0 | Ng details how AI coding tools make rapid prototyping 10x faster and advises writing throwaway, insecure code during early exploration. | |
| AI Coding Assistance and Shifting Software Engineering Philosophies | 0 | 0 | 0 | 0 | Ng explores shifting engineering mindsets, where code is no longer a precious artifact and stack choices become two-way doors. | |
| Empowering Everyone to Build with AI | 0 | 0 | 2 | 0 | Ng rejects conventional advice that AI will replace the need to learn coding, arguing instead that everyone across all job roles should code. | |
| Product Feedback Tactics and Shifting Eng/PM Ratios | 0 | 0 | 0 | 0 | Ng explains why product management and customer feedback loops have become the new bottleneck as engineering velocity increases. | |
| Technical Understanding and GenAI Building Blocks | 0 | 0 | 0 | 0 | Ng breaks down why technical understanding of GenAI building blocks is essential for compounding combinatorial advantage. | |
| Conclusion: Building at Speed with AI | 0 | 0 | 0 | 0 | Ng summarizes the core tenets of his talk before taking questions from the audience. | |
| Q&A: The Future of Compute and Addressing Hype Narratives | 0 | 5 | 4 | 0 | Ng pushes back strongly against hype narratives around space GPUs, nuclear-only AI power, and apocalyptic extinction scenarios. | |
| Q&A: Responsible AI vs. Overhyped Safety Narratives | 0 | 4 | 3 | 0 | Ng reframes the 'AI safety' debate into 'responsible AI', comparing AI technology to an electric motor where safety depends on application. | |
| Q&A: Startup Moats and Building Products Users Love | 0 | 3 | 1 | 0 | Ng advises an aspiring founder to prioritize product-market love over premature obsession with defensible moats. | |
| Q&A: Agentic Accumulation, Token Costs, and Model Flexibility | 0 | 4 | 2 | 0 | Ng addresses technical questions regarding token costs and agentic accumulation, advising founders not to worry prematurely about token bills. | |
| Q&A: How AI is Transforming the Educational Landscape | 0 | 5 | 4 | 0 | Ng concludes by addressing educational transformation and passionately attacking regulatory lobbying (like SB 1047) aimed at shutting down open source. |
Statements from this episode (33)
Ng: AI Fund builds an average of one startup per month
“AI Fund's a venture studio, and we build an average of about one startup per month”
Ng: The biggest AI economic opportunities must be at the application layer
“And even though a lot of the PR excitement and hype has been on these technology layers, it turns out that almost by definition, the biggest opportunities have to be at the application layer. Because we actually need the applications to generate even more reve…”
Ng: Agentic AI is the most important technology trend in artificial intelligence
“And in terms of AI tech trends, if you ask me what's the most important tech trend in AI, I would say Is the rise of agentic AI.”
Ng: Agentic workflows determine whether AI applications work or fail
“So for a lot of the projects that AI fund has worked on, everything from pulling out complex compliance documents, to medical diagnosis, to reasoning about complex legal documents, we found that these agentic workflows are really a huge difference between work…”
Ng: Subject matter expert intuition beats slow data for startup decisions
“Getting data for a lot of startups is a slow mechanism for making decisions. And a subject matter expert with a good gut is often a much better mechanism for making a speedy decision.”
Ng: Over-pivoting on customer feedback reveals a lack of domain knowledge
“If every time you talk to a customer, you totally change your mind, probably means you don't know enough about that sector yet to have a really high quality concrete idea”
Ng: AI coding makes prototyping 10x faster, production only 50% faster
“When writing production quality code, maybe we're at 30 to 50% faster with AI systems. Hard to find a rigorous number. Maybe. Be as plausible to me. But in terms of building quick and dirty prototypes, we're not 50% faster. I think we're easily 10 times faster…”
Ng: Engineering teams should write insecure code for local prototypes
“I routinely go to my teams and say, go ahead, write insecure code, because if this software is only going to run on your laptop, and if you don't plan to maliciously hack your own laptop, It's fine to have insecure code, right? But of course, after it seems to…”
Ng: Falling behind on AI tools causes a severe productivity gap
“And the interesting thing is, if you're even half a generation or one generation behind, it actually makes a big difference compared to if you're on top of the latest tools.”
Ng: Code itself is becoming much less valuable as engineering costs drop
“One surprising thing is, we're used to thinking of code as this really valuable artifact because it's so hard to create, but because the cost of software engineering is going down, code is much less of a valuable artifact as it used to.”
Ng's teams rebuilt a codebase three times in one month using AI
“So I'm on teams where, you know, we've completely rebuilt the code base three times in the last month, right? Because it's not that hard anymore to just completely rebuild the code base, pick a new data schemer, it's fine, because the cost of doing that has pl…”
Ng: AI turns software architecture decisions into reversible two-way doors
“So choosing the software architecture of your tech stack used to be a one-way door. Once you're built on top of a certain tech stack, you know, set a database schema, really hard to change it. So that used to be a one-way door. I don't want to say it's totally…”
Ng: Advising people not to learn to code is terrible career advice
“Over the last year, a bunch of people have advised others not to learn to code on the ground so AI will automate it. I think we'll look back on this as some of the worst career advice ever given, because as better tools make software engineering easier, more p…”
Ng: Employees across every job role should learn how to code
“I have a controversial opinion, which is, ah, I think it's time for everyone of every job role to learn to code, and in fact, on my team, you know, my CFO, my head of talent, my recruiters, my front desk person, all of them know how to code, and I actually see…”
Ng: Product management is becoming the software bottleneck as engineering accelerates
“With software engineering becoming much faster, the other interesting dynamic I'm seeing is that the product management work, getting user feedback, deciding what features to build, that is increasingly the bottleneck.”
Ng: One project team proposed staffing twice as many PMs as engineers
“So literally yesterday, one of my teams came to me, and for the first time, when we're planning a head comfort project, this team proposed to me not to have one PM to four engineers, but to have one PM to 0.5 engineers. So the team actually proposed to me. I s…”
Ng: A/B testing is now one of the slowest product feedback tactics
“And I know Silicon Valley, you know, we like to talk about A-B testing. Of course, I do a ton of A-B testing, but contrary to what many people think, A-B testing is now one of the slowest tactics in my menu, because it's just slow to ship it.”
Ng: Use A/B testing to hone gut instincts, not just pick variants
“I don't just use the result of A-B test to pick product A or product B. My team will often sit down and look carefully at the data to hone our instincts, to speed up, to improve the rate at which we're able to use the first tactic to make high quality decision…”
Ng: Bad AI architecture decisions slow startups down by 10x, not 2x
“It feels like one bit of information can at most buy you a two-way speed up. And I think in some theoretical sense, that is true. But what I see in practice, if you flip the wrong bit, you're not twice as slow. You spend like 10 times longer chasing a blind al…”
Ng: AGI is overhyped; humans will outperform AI for a long time
“I feel like AGI has been overhyped.
And so for a long time, there'll be a lot of things that humans can do that AI cannot.”
Ng: AI existential risk claims are ridiculous hype used to aid fundraising
“This idea that AI is so powerful, we might accidentally lead to human extinction. That's just ridiculous, but it is a hype narrative that made certain businesses look more powerful, and it got Yeah, ramped up and actually help certain businesses fundraising go…”
Ng: New foundation models will not casually wipe out thousands of startups
“We're so powerful that by training a new model, we will casually wipe out thousands of startups. That's just not true. Yes, Jasper ran into trouble, small number of companies got wiped out, but it's not that easy to casually wipe out Thousands of startups.”
Ng: AI compute does not exclusively require nuclear power
“AI needs so much electricity, only nuclear power is good enough for that. You know, that wind, solar stuff, not, this is not true.”
Ng: AI safety is a function of the application, not the technology
“Safety is not a function of technology. It's a function of how we apply it. So like electric motor, you know, You can't, the maker of electric motor can't guarantee that no one will ever use it for an unsafe downstream task. Like using electric, electric motor…”
Ng: Sensationalized AI danger hype is being weaponized against open-source software
“And I feel like this has been used as a weapon against open source software as well, right? Which is really unfortunate.”
Ng: Startup moats are overhyped and evolve after launching a loved product
“I find that moats tend to be overhyped, actually. I find that more businesses tend to start up with a product and then evolve eventually into a moat.”
Ng: AI application opportunities far exceed the number of skilled builders
“The number of opportunities, meaning the amount of stuff that is possible that no one's built yet in the world, seems much greater than the number of people with the skill to build them. So definitely at the application layer, it feels like there's a lot of wh…”
Ng: Early-stage startups should not worry about API token costs
“My most common advice to developers is to first approximation. Just don't worry about how much tokens cost. Only a small number of startups are lucky enough to have users use so much of your product that the cost of tokens becomes a problem.”
Ng: Switching foundation models has lower friction than switching orchestration platforms
“So it turns out that switching costs for foundation models is relatively low, and we often architect our software. Oh, AI Suite is open sourcing that my friends and I worked on. So it makes switching easier. Switching costs for the orchestration platforms is a…”
Ng: Tech will spend a decade mapping work to agentic workflows
“And for the next decade, we'll be looking at the work that needs to be done and figuring out how to map it to agentic workflows.”
Ng: AI Fund has killed economically viable projects on ethical grounds
“But AI Fund, we've killed multiple projects, not on financial grounds, but on ethical grounds where there are multiple projects. We looked at the economic case is very solid, but we said, you know what, we don't want this to exist in the world, and we just kil…”
Ng: California's SB 1047 would have harmed open-source AI without improving safety
“So I think typing up dangers, supposed false dangers of AI, in order to get regulators to pass laws, like the proposed SB-SB-Ten-Forty-Seven in California, which, thank goodness, we shut down, Would have put in place really burdensome regulatory requirements t…”