Chapters (Powered by ChapterMe) - 00:00 - YC Spring Batch 00:17 - Coming Up 01:07 - Intro: YC hosted a 300-person AI alumni retreat. 01:48 - Startups average 10% weekly growth. 02:34 - One company hit $12M ARR in a year. 03:04 - Startups achieve $1M ARR in six months. 04:54 - Enterprises embrace AI, driving huge demand. 06:00 - Make something people want 07:04 - Technical complexity limits AI competition. 07:50 - Eval sets are key assets for AI firms. 08:42 - Prompting and testing drive AI success. 09:31 - Designers use AI to generate code directly. 10:09 - AI speeds up product iteration and creation. 10:56 - AI automates tasks, creating wealth. 13:32 - AI could universalize quality medical care. 14:27 - Economies may balance AI and human roles. 15:21 - AI can democratize life-coaching tools. 16:06 - AI boosts living standards by automating tasks. 17:09 - Positive AI development fosters creativity. 19:21 - Predictive models ensure safe AI growth. 21:01 - Human oversight in AI remains critical. 22:11 - OpenAI fosters competition in AI markets. 24:48 - Early adopters shift from Google to AI tools. 26:13 - Cursor gains fast adoption among YC startups. 28:02 - AI tools reduce reliance on SaaS software. 29:20 - AI enables scaling with fewer resources. 31:22 - Lean growth replaces blitz scaling. 33:06 - AI pricing ties directly to ROI. 34:08 - Startups excel by rapidly adapting tools. 36:27 - Cutting-edge tech drives startup success. 38:20 - Vibe: YC empowers small, ambitious AI teams. 39:13 - Outro
I think AI is the most fascinating thing that's happening in the world today, but there is just so much to learn, that's why I started my own channel "The AI Oldtimer" to chronicle my journey of discovery...!
@@antonyba9 delusional* But no.. they aren't. I find I move quicker when I build MVPs by myself. To have full context of a project, not have to communicate with someone constantly, etc. allows me to move much quicker. And with the help of AI.. I've never built things this fast in my life! Although, I would argue that the importance of a business/sales co-founder is important. Building is only half.
@@antonyba9 Co-founder dispute are the #1 reason of startup death. Then if the startup does not die, one of the co-founder usually takes the helm anyways : Steve Jobs over Wozniak, Bill Gates over Allen, etc. The whole co-founder advice is just stupid.
19:50 LLMs are trained on data generated by humans, and human-generated content can reflect survival instincts,. So these patterns may appear in LLM outputs. even though LLMs are simply predicting the next token, their predictions can reflect these embedded survival-oriented patterns.
As a software engineer who's dedicated their career to solving problems and writing great code I've recently been straddling the line between build an ai powered startup and "lay down on a bed" I was slightly leaning towards lay down but this video was inspiring and helpful.
We are missing one thing, here about having less employees. Old days people use to make software using assembly and mainframes. We thought and scared that using PCs will make developers productive so companies will need only 10 devs instead of hundreds. Turns out, you still need 100 devs, because complexity of systems increases following the tools. So image how much cool stuff can develop 100 developer vs 10 devs. Work will always be there
And technically the people speaking on computers regarding the threat to jobs, could still be right, no one said when it would happen but it seems to be following the motions right now.
AGI is so different from anything we’ve seen, it’s not intended to be a tool but to actually mimic the human worker not supplement it. Whether it’s a threat to jobs really depends on how much autonomy will be baked into AGI. Will it need to be continuously promoted to function or does it continuously carry out tasks without the need for repeat prompting?
It would be so interesting to hear an episode about the AI economy. A lot of people already understand and accept the exponential growth of AI in our daily lives - ideas like those of Ray Kurzweil are becoming more mainstream. But when it comes to the economy and job loss, that same understanding of exponential growth seems to disappear. It’s not so much that jobs will completely vanish, but how quickly people and industries will have to adapt, relocate, and shift jobs from one market to another (if there will be any). It would be great if this could be discussed in one of the episodes.
We couldn't agree more! After a couple months in private beta, we just launched 2 weeks ago. It's been a thrill, helping the Unheard-99% start podcasting!
He said that this guy has skipped using Figma at all. Having the sense of design I am building products directly in code for 10+ years. So it’s not because LLMs are here. I think that Figma is good for teams where some people could code but not able to design and others could design but can’t code. I can code and design so the Figma is not necessary for me even for quick prototyping:)
Has anyone come across that not all AI agents are giving optimal outcome for the prompts you put ? Still lot needs to be refined in image and video generation.
Friend of mine build a really great product and is trying to launch it. But customers want AI, instead of adding it he’s adamant AI is a fad and a waste of time. Dude is a high level soft Eng and I can’t get him to see the change over the last year…
The industry primes are going to gobble up a huge amount of the big industry ai capabilities. The places to look for small players will be in the cracks. There’s money to be made gluing up the non trivial agentic interfaces. Don’t sleep on it.
UBI won’t work - Universal Basic Services addresses the issues by ensuring that people have a meaningful standard of living, that isn’t inflation dependent. There is a costed model by University College London. In a post AI economy we need to move the focus away from the means (money) and toward the outcome (wellbeing). It doesn’t matter if people get an allowance of £100k a year, if they can’t secure anything meaningful for their welfare. Additionally the bottleneck in the post AGI future is distribution of useful services. We already create life saving drugs and price people out of having them. The Lyme disease vaccine was pulled due to profit lability. Aside from maybe the price of production might lower, there is no structural guarantee cheaper production = higher distribution vs better margins.
Enterprise agents will be inferior to personally owned agents with access to all your data, building abilities for personal AI is the real marketplace, find out more at Kwaai, (and let these folk chase easy money with small visions)
Love the eval_set - coming from an age where "setq", cons, car, cdddddddddddr, metacircular evaluations, and all the good jazz was on curriculum, Day 1 basically. :D
Depends on what you want.. if you want to start a business then 99% of the work is find the product idea. Then you can asknwhat steps(including skills) would be necessary tonacchieve your product or service. See skill as a "step" rather than the goal
Chapters (Powered by ChapterMe) -
00:00 - YC Spring Batch
00:17 - Coming Up
01:07 - Intro: YC hosted a 300-person AI alumni retreat.
01:48 - Startups average 10% weekly growth.
02:34 - One company hit $12M ARR in a year.
03:04 - Startups achieve $1M ARR in six months.
04:54 - Enterprises embrace AI, driving huge demand.
06:00 - Make something people want
07:04 - Technical complexity limits AI competition.
07:50 - Eval sets are key assets for AI firms.
08:42 - Prompting and testing drive AI success.
09:31 - Designers use AI to generate code directly.
10:09 - AI speeds up product iteration and creation.
10:56 - AI automates tasks, creating wealth.
13:32 - AI could universalize quality medical care.
14:27 - Economies may balance AI and human roles.
15:21 - AI can democratize life-coaching tools.
16:06 - AI boosts living standards by automating tasks.
17:09 - Positive AI development fosters creativity.
19:21 - Predictive models ensure safe AI growth.
21:01 - Human oversight in AI remains critical.
22:11 - OpenAI fosters competition in AI markets.
24:48 - Early adopters shift from Google to AI tools.
26:13 - Cursor gains fast adoption among YC startups.
28:02 - AI tools reduce reliance on SaaS software.
29:20 - AI enables scaling with fewer resources.
31:22 - Lean growth replaces blitz scaling.
33:06 - AI pricing ties directly to ROI.
34:08 - Startups excel by rapidly adapting tools.
36:27 - Cutting-edge tech drives startup success.
38:20 - Vibe: YC empowers small, ambitious AI teams.
39:13 - Outro
The chemistry and vibe between the original 4 of this group is so amazing..they literally understand each other inside out
I think AI is the most fascinating thing that's happening in the world today, but there is just so much to learn, that's why I started my own channel "The AI Oldtimer" to chronicle my journey of discovery...!
AI enables me to bypass the ritual YC advice : "Get a co-founder"
really
You're dillutional. Image you are making something alone or with someone (image ideal partner ), both using ai? Result will be complete different.
@@antonyba9 delusional* But no.. they aren't. I find I move quicker when I build MVPs by myself. To have full context of a project, not have to communicate with someone constantly, etc. allows me to move much quicker. And with the help of AI.. I've never built things this fast in my life! Although, I would argue that the importance of a business/sales co-founder is important. Building is only half.
@@antonyba9 Co-founder dispute are the #1 reason of startup death. Then if the startup does not die, one of the co-founder usually takes the helm anyways : Steve Jobs over Wozniak, Bill Gates over Allen, etc. The whole co-founder advice is just stupid.
If you don’t have a cofounder there isn’t going to be any funding for your startup. And the start will fail
19:50 LLMs are trained on data generated by humans, and human-generated content can reflect survival instincts,. So these patterns may appear in LLM outputs. even though LLMs are simply predicting the next token, their predictions can reflect these embedded survival-oriented patterns.
Deepseek heavily used synthetic data!
As a software engineer who's dedicated their career to solving problems and writing great code I've recently been straddling the line between build an ai powered startup and "lay down on a bed"
I was slightly leaning towards lay down but this video was inspiring and helpful.
We are missing one thing, here about having less employees. Old days people use to make software using assembly and mainframes. We thought and scared that using PCs will make developers productive so companies will need only 10 devs instead of hundreds. Turns out, you still need 100 devs, because complexity of systems increases following the tools. So image how much cool stuff can develop 100 developer vs 10 devs. Work will always be there
but it will be 100 ai devs, not 100 human devs.
And technically the people speaking on computers regarding the threat to jobs, could still be right, no one said when it would happen but it seems to be following the motions right now.
AGI is so different from anything we’ve seen, it’s not intended to be a tool but to actually mimic the human worker not supplement it.
Whether it’s a threat to jobs really depends on how much autonomy will be baked into AGI. Will it need to be continuously promoted to function or does it continuously carry out tasks without the need for repeat prompting?
Paul knows a lot! Lightcone is the best podcast in the startup world!
Whats the use? Isn't OpenAI, MS, Google, Meta, AGI going to build it anyways and you have to close your startup?
Because individual human agency is amplified far more today than ever before. -Garry
@@ycombinatorbuilding tailored solutions for a target demographic and create a bunch of content. None of my friends use AI yet, still early
Perplexity proves you dont have to close your startup
@@ycombinatorappreciate the response. Please could you elaborate?
YC just funded a startup, CopyCat, and OpenAI's operator is almost already there.
Not even a memtion of Deepseek R1 guy's, it proves smaller more nimbler startups can compete with the massive tech giants.
We are working on a video about this
It would be so interesting to hear an episode about the AI economy. A lot of people already understand and accept the exponential growth of AI in our daily lives - ideas like those of Ray Kurzweil are becoming more mainstream. But when it comes to the economy and job loss, that same understanding of exponential growth seems to disappear. It’s not so much that jobs will completely vanish, but how quickly people and industries will have to adapt, relocate, and shift jobs from one market to another (if there will be any). It would be great if this could be discussed in one of the episodes.
We couldn't agree more! After a couple months in private beta, we just launched 2 weeks ago. It's been a thrill, helping the Unheard-99% start podcasting!
Learn, build and sell.
Applied AI and learning systems will save our day” 😊
This is inspiring. Makes me want to apply.
Don't. They dont want you.
HA! Im the introvert with AI ... I've been waiting on this for 20 years...
What's the name of the tool they suggest to use instead of Figma? Minute 09:30
Claud LLM
Antropic Claude
I use v0 from Vercel and as of today ChatGPT o1 and Canvas
He said that this guy has skipped using Figma at all. Having the sense of design I am building products directly in code for 10+ years. So it’s not because LLMs are here. I think that Figma is good for teams where some people could code but not able to design and others could design but can’t code. I can code and design so the Figma is not necessary for me even for quick prototyping:)
They are using Replit
Shoutout to the presenters for an interesting conversation while balanced on rattan bins.
What is the name of the company that achieved $12 million ARR in a year?
I am a CSE AIML student from India, am I eligible to apply? I have a few great startup ideas and projects. Thanks
Yes
Has anyone come across that not all AI agents are giving optimal outcome for the prompts you put ? Still lot needs to be refined in image and video generation.
Friend of mine build a really great product and is trying to launch it. But customers want AI, instead of adding it he’s adamant AI is a fad and a waste of time. Dude is a high level soft Eng and I can’t get him to see the change over the last year…
The industry primes are going to gobble up a huge amount of the big industry ai capabilities.
The places to look for small players will be in the cracks.
There’s money to be made gluing up the non trivial agentic interfaces. Don’t sleep on it.
Lightning startup podcasts 🍿
OpenAI used Google’s research and simply scaled it, but hasn’t built AGI or its tech at all
Why don't you simply scale?
UBI won’t work - Universal Basic Services addresses the issues by ensuring that people have a meaningful standard of living, that isn’t inflation dependent. There is a costed model by University College London. In a post AI economy we need to move the focus away from the means (money) and toward the outcome (wellbeing). It doesn’t matter if people get an allowance of £100k a year, if they can’t secure anything meaningful for their welfare. Additionally the bottleneck in the post AGI future is distribution of useful services.
We already create life saving drugs and price people out of having them. The Lyme disease vaccine was pulled due to profit lability.
Aside from maybe the price of production might lower, there is no structural guarantee cheaper production = higher distribution vs better margins.
Enterprise agents will be inferior to personally owned agents with access to all your data, building abilities for personal AI is the real marketplace, find out more at Kwaai, (and let these folk chase easy money with small visions)
Love the eval_set - coming from an age where "setq", cons, car, cdddddddddddr, metacircular evaluations, and all the good jazz was on curriculum, Day 1 basically. :D
When did you visit Vietnam Garry? Next time, if possible hit me up, I would like to invite you for lunch
harj is having a zuck arc lol, bulking up and deeper voice
Here 👋🏾
Whats the best skill to know tho.
Depends on what you want.. if you want to start a business then 99% of the work is find the product idea. Then you can asknwhat steps(including skills) would be necessary tonacchieve your product or service. See skill as a "step" rather than the goal
Been reaching out to them for a while we built Agi and asi but I guess it’s not something they are interested in.
the best time to start an AI startup was like 5 years ago. nowadays it's already a oversaturated market.
False, will age poorly sorry
These videos in 1080P absolutely suck tbh.
why is that your focus and main concern? It's a podcast just listen and learn.
What’s UBI? 😅
Universal basic income
i got something good, beutiful and useful. Can we talk?
🙏
What's his name? Showing off his intellect.
The guy with the watch.
Please do a video on the type of tech starts ups to start.
Talk about monetization models.
Thanks.