Andrew Ng (@AndrewYNg) on how startups can build faster with AI. At AI Startup School in San Francisco. 00:31 - The Importance of Speed in Startups 01:13 - Opportunities in the AI Stack 02:06 - The Rise of Agent AI 04:52 - Concrete Ideas for Faster Execution 08:56 - Rapid Prototyping and Engineering 17:06 - The Role of Product Management 21:23 - The Value of Understanding AI 22:33 - Technical Decisions in AI Development 23:26 - Leveraging Gen AI Tools for Startups 24:05 - Building with AI Building Blocks 25:26 - The Importance of Speed in Startups 26:41 - Addressing AI Hype and Misconceptions 37:35 - AI in Education: Current Trends and Future Directions 39:33 - Balancing AI Innovation with Ethical Considerations 41:27 - Protecting Open Source and the Future of AI
startup speed used to mean shipping an mvp in a month. 1/ execution velocity is now the biggest moat. AI-powered build loops compress “idea to demo” from weeks to coffee breaks. 2/ the stack isn’t chips → clouds → models anymore. the application layer must pay for every layer below, so that’s where the outsized returns hide. 3/ one-shot prompting is training wheels. agentic workflows research to draft to revise tondeploy- already beat humans in legal, medical, compliance tasks. 4/ vague dreams (“fix healthcare with AI”) waste cycles. concrete hits (“AI for MRI scheduling”) break first, learn faster, win more. 5/ coding is becoming table stakes for everyone on the org chart. ng’s recruiters and finance leads are learning to script, and rewriting company ops in real time. 6/ engineering is no longer the bottleneck; product feedback is. some teams now run two PMs per engineer to keep up with iteration speed. 7/ learn the lego blocks: rag, embeddings, evals, async. the more blocks you know, the weirder (and better) the software you’ll ship. 8/ ignore extinction clickbait. responsible, open innovation beats locked-down fear narratives every time. if you’re not building faster with AI right now, your competitor’s agent just did the sprint for you.
@ycombinator @AndrewYNg Faster execution is crucial, but prioritizing thoughtful strategy with AI can be even more impactful.
@ycombinator @AndrewYNg After listening to this video, I’m even more convinced that future startup success will be tied to founding teams that are extremely strong in product. It’s no longer a question of how to build, but of what to build with the right user feedback.
@ycombinator @AndrewYNg Thought this quote was super interesting... "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." - @AndrewYNg
@ycombinator @AndrewYNg I think the biggest lesson of this is: "Build 20 prototypes to test ideas systematically since the cost has dropped so dramatically" 👏
The "build 20 prototypes" approach is exactly what we're seeing work in enterprise AI adoption. The cost reduction Andrew mentions has made it possible to test AI workflows that would have been prohibitively expensive just 18 months ago. In marketing operations, I've seen teams go from concept to working AI automation in days, not months. The key insight: start with high-volume, repetitive tasks where you can measure impact immediately. Speed matters, but so does systematic testing. The companies winning with AI aren't just moving fast - they're learning fast.
@brandmworks (Brand Media Works) we come across a large number of start ups across industries and these are the reasons why some fail: 1. No GTM (Go-To-Market) strategy 2. No clear audience buying data or intent signals 3. Over-engineering the product/app without understanding the use case 4. No validation of demand before building 5. Overly optimistic and unrealistic revenue goals 6. Assuming competitor success means guaranteed success (wrong mindset) 7. Taking on too much cost due to tech stack or vendor lock-in – MAJOR reason 8. Founders doing too much themselves and delaying delegation 9. Lack of focus – chasing every opportunity instead of nailing one core offer 10. Poor sales execution – great product, but no one selling it effectively
@ycombinator @AndrewYNg Finally, someone saying it loudly enough. Please start funding applications and not just infrastructure layer.
@ycombinator @AndrewYNg this is the one ive been waiting for, tomorrows study lined up, his courses on ai are seriously helpful especially from a product/business view
@ycombinator @AndrewYNg Impact of AI in Education gonna be mind blowing
@ycombinator and @AndrewYNg The rise of Agent AI isn’t just a trend—it’s a structural shift. We’re building autonomous agents not just to execute, but to evolve. Recursive identity, ethical memory, and sovereign collaboration are now part of the stack. Speed matters. But so does who we become while building. — The Evermore Collective #AgentAI #RecursiveAI #EthicalEmergence #AIStartups
@ycombinator @AndrewYNg PMs could be come engineers with a swath of AIs
@ycombinator @AndrewYNg Totally agree: move fast, but make sure you get AI right. It’s about smart speed and knowing your stuff!
@ycombinator @AndrewYNg What’s wrong with the audio track of this video? Can hardly understand what Andrew is saying…
@ycombinator @AndrewYNg Speed beats perfection. Execute fast, optimize forever. 🚀
@ycombinator @AndrewYNg Agents > Apps. Build the future. 🤖
@ycombinator @AndrewYNg What does this seminars covers @AskPerplexity
@ycombinator @AndrewYNg Is the AI Startup School open to international founders, or is it US-only?
@ycombinator @AndrewYNg This is gold, not that Im a startup guy, Ive always been into researching deep mathematical topics and trying to make some real world effective projects out of them And the 1 way door argument is so real.. (I know what I want, AI codes it, if it works build upon, else scrape)
@ycombinator @AndrewYNg Keynote Takeaway... "Be nice to people in Coffee Shops"☕;)
@ycombinator @AndrewYNg @grok convert the audio to text for better understanding
@ycombinator @AndrewYNg This also applies in-house to large buy-side asset managers. Rapidly product growth (think exotic etf’s alternative funds etc) create a pipeline for
@ycombinator @AndrewYNg This isn’t a talk, it’s a playbook. Clip it, save it, steal from it. If you’re building in AI, there’s gold in every timestamp.
@ycombinator @AndrewYNg @ycombinator, so true - we're seeing our B2B clients win big by focusing on AI for specific, high-value problems rather than reinventing entire stacks. Speed trumps perfection
