Chapters (Powered by ChapterMe) - 00:00 - Coming up 00:47 - Intro 01:15 - Making an app with Replit 06:19 - Feel the AGI, personal software era 08:07 - Having AI code the way humans do 09:51 - You should still learn to code! 11:42 - The underlying tech 17:19 - The path to AGI 19:41 - What users made with Replit 25:56 - Challenges in resetting the org 33:29 - Future plans 36:12 - Outro
But they mentioned in the video, that it's still not a replacement for you taking responsibility of the code, yet. Probably it's only a great tool to get started with a project
As someone whose business's flagship app was built in Replit with help from LLMs, Amjad is way too optimistic. For anything more sophisticated than toy apps, you've got to have software architecting, system and database administering, and lots of software engineering expertise just to pose the right questions to the AI. We're getting there, but it will take at least a couple more years to get where Amjad describes as today.
As a software developer i find the title of this video surreal. IT IS JUST NOT TRUE THAT YOU CAN DEVELOP A SOFTWARE PRODUCT (EVEN A VERY SIMPLE ONE) AND DEPLOY IT WITH JUST AI.
Yes you can, I made two Apps on AppStore with 99% ClaudeSonnet 3.5 and Cursor AI. And much more. Feels like I have a 10 person team delivering code for me in 10 seconds, that other does in 10 days
As someone who has done product design for the past 30 years, I didn't know how to code, but I did know how to communicate with programmers what exactly I wanted to do. Now I no longer have to rely on other people to realize my ideas. And let me tell you AIs are lot easier to communicate that many humans I had to deal with in the past years…
I used Replit for the first time Yesterday and it Started spinning and got stock in some loop so had to stop it and redo some times o but Eventually it did Create my first LLM API integration and deployed it which made me really happy
I haven't coded in almost 3 decades. Recently, I started using LLMs to write code. It's been fun, I mean REALLY fun because I remembered the days of coding freestyle. However, it has been frustrating too. There is no doubt that the more you know about the language, the faster it goes. I've spent 10+ hours trying to get something to work because of the limitations of LLMs and of course me. I have to prompt the LLMs correctly to reduce the development time and the coding errors. Also, I have to use multiple LLMs to accomplish the task. In my experience, the coding quality is in this order GPT o1 > Claude 3.5 > Gemini > Llama 3.1. If I get stuck in one, I switch to the other to fix the problem. Regardless, this beats learning a language when I just don't have time.
This comment describes my experience exactly. Plus one other point: LLMs are good at writing code and terrible at debugging. Their debugging approach consists of (1) Let's add a ton of logging to find out what is going on, and (2) Let's imagine some strange edge-case scenarios that have nothing to do with reality and code around them with a lot of useless junk. So, even if you don't want to learn a new language, you have to use your development experience to help it find many simple logic bugs. As Harold says, it can still be fun, and I have created extremely useful stuff. Prime example is Apps Scripts for Google Sheets. I am a very experienced spreadsheet user, but I had never created Apps Sheet code before. I am thrilled with what I have been able to do with AI. The scripts are far from being an application, but they can leverage the power of spreadsheets and other tools like Looker Studio to do very useful things.
News flash: Everyone has been able to code for a while now, even before LLMs. There's been a nearly limitless amount of resources to learn coding, many of them for free or at a negligible cost. Not many people do because it takes determination and work to become decent at it. Everyone is looking at LLMs as a free lunch to bypass that work, but code is still being produced by it, and it will still need to be understood if you ever want to create any serious apps.
@@thebicycleman8062 I don’t delete comments. If you think you can produce millions of lines of code in a production setting without needing to at least verify them before deployment, I’ve got bad news for you. Bugs and unintended behavior can cost you money and customers in the real world.
This saves me like 5 second compared to just using Cursor AI that have been out for several months. The drawback is lock-in to replit, compared to Cursor AI that is just a good IDE over Claude Sonnet 3.5 and ChatGPT o1
Sounds true, but everyone can talk; yet not everyone is a public speaker. Quality, choosing what to build and how to deliver it still will be the differentiators (*even when eventually coding is a fringe role)
@@IntegrandoIA I agree with you. However choosing what to build and how to deliver eventually will not be a topic too. AI will tell you what is the most effective strategy. So how would a business stand out?
Everyday people say everyone can now code but I am yet to see a single app that a non-developer has built with only LLM. Instead, my workload on upwork keeps climbing. Are these clients not aware they can build their apps with just LLM?😅
This is an excellent roundtable discussion! Note that one of the speakers is talking about the difference between effective AGI and true AGI. It has become quite clear to many of us AI aficionados that effective AGI is just “a few thousand days away”
I have aged out as a coder, but system architecture and understanding of data is an enduring skill. I would love to build a real application with AI, something with a database, user classes, interactive screens where users can add data, interact with data, and report data. One after another, the "application" examples in these videos are just wrappers around an AI conversation. That is not an application. It's a toy.
@@SteveKowarsky for me when i said efficiency that includes functional requirments - so efficiency in the sense that it effiiently functions - so yes these new things are not toys anymore, the problem is 99% of A.I critics that come from software dev background of any 'shade' have a personal bias to never want it to work out for obvious conflict of interests, and the problem is this does nothing but just keep their eyes shut down to the extremely rapidly growing oppurtunities and new ways to evolve with the field. How about this Steve, tell me in brief a certain app functinoality or even a tech stack you would consider "a reall application" as opposed to a "toy" and lets have an honest conversation and I will tell you if I have managed to create something similar to it or one of my A.I-Based Coders - Coz I am in a community of a large large group of A.I-Based Developers and you would be surprized what we have managed to build. - I am genuinely interested in knowing what a developer with experience prior to 2023(LLM Widespread Integration) views matters on development in general, I could probably learn a lot from you coz I am definitely aware as an A.I-Based Coder of only 8-month experience that a lot of fundamentals I definitely lack coz I started coding around 8 month ago not even knowing what a code even is, I just thought it was like some hacker matrix thing
AI driven agent-assistant will make engineers life less stressful and also help to spot a bug in a codebase in minutes instead of scrolling through hours
Commenters, the direction is more important than the snapshot. I used it on day 1 and was impressed but it wasn’t fully baked. But it’s a work in progress and this is going to learn faster than most people will. Make your decisions with the assumption that these tools get better every year. They might not out-code a top coder, bit a top coder with it is better than one without. And many businesses don’t have top coders.
Sounds like solutionism. You sure can use these AI Agents to code simple app that will work half the time when you give em the right inputs. Going from a half baked app to a production ready one is where it gets really challenging. And, the bar is so so so high now so I wouldn't say AI agents can build your 'whole' app.
@@ycombinator I agree that current AI tools are useful but often too general-purpose, which limits how deeply they can dive into specific domains-especially when it comes to coding. Even the specialized coding AIs out there still face these limitations. That’s why we’ve taken a different approach, and we’re launching our solution in just 2 weeks. Our AI agents are designed specifically for backend programming, and we’re already in the testing phase. Unlike most solutions, ours are vertically built, and in certain areas, we orchestrate our AI agents in a more opinionated way. While this can occasionally limit creativity, it ensures that we’re building systems that actually work. We believe the future of solving every software problem won’t come from one tool but rather the union of specialized systems like ours, each focused on specific areas of development." ı will leave a link here soon
On top of drawing in the UI, it would be nice to be able to take pictures of something sketched on a piece of paper. It's so much easier to use pen and paper as compared to freeform drawing on a screen.
Insights By "YouSum Live" 00:00:30 Personal software revolutionizes app development 00:00:34 Users can build apps in minutes 00:01:30 Live demo showcases mood tracking app 00:08:17 AI assists in coding and debugging 00:09:20 Incremental learning is key for coding 00:20:27 AI tools enhance creativity and productivity 00:22:58 Users create apps faster than ever 00:34:35 Future improvements focus on reliability 00:35:01 Enhanced user interaction through drawing 00:36:01 Advanced users gain more control over code Insights By "YouSum Live"
A 15 USD/month for Replit Core for using an AGI precursor, it doesn't seem a bad deal when you can play and build a ton. Team AI Aegnt Revolution 2025!
Hi, please do another podcast about Vertical Ai Agents - Last one was great! You could showcase some other SAAS start ups, that are using AI to leverage their business model. Thank you!
So many people commenting are shockingly critical and pessomistic. It's like seeing the reaction to cameras being added to a flip-mobile-phone back in 1999. Of course it's crap in its infancy. What they are failing to see is how fast we got here, and where AI could be in mere 5-10 year horizon.
We are launching our MVP in 2 weeks, and it creates complex apps that actually work. I will edit here and provide a link 🚀🚀🚀(u still need to know how to code)
@@0x0007Nexcept he's right. There's plenty of studies showing that code bases made with AI tend have security issues, a lot of bugs, the majority of commits are just bug fixes for ai generated code and it takes longer to fix as opposed to just writing without AI, leading to a lot of tech debt. If you don't have a technical background, you can't even perceive the flaws in your code
Everyone can code. What that means at this time is everyone can code simple apps with a minimal outlay of time and knowledge. I now use dozens of personalized apps I generated with AI, often completed before breakfast. Where I might have spent an afternoon making a spreadsheet, I now spend ten to twenty minutes prompting an AI. It's often faster and more reliable that trying to search for an app online that is full of SEO, ads, and unnecessary submit buttons that reload the page for more ads. Real devs can build apps with user logins, persistent state, and lots of features. But normies like me can now build a simple app that uses local storage, import and export buttons, and the exact features they need. There are so many people who sit around wishing someone would build an app that does one specific thing they need, but the apps either don't exist or the features are made for a broad market. Tools like this can serve those individuals.
The messaging here seems a bit confusing- on the one hand 'anyone' can code, which implies that we will all be interacting on the level of natural language- but there also seems to be a view that learning to code will still be required? Given the entire history of programming computers is a story of increasing levels of abstraction away from the need to 'speak the computer's language' then isn't the long term outcome of this trend inevitably going to be a programming languge that everyone does already know, which is spoken and written language as used in everyday life?
Neither! PWA is not simple. A.I can outline how to build a pwa very well and concise but it cannot build it for you hence you still have to know how to write code.
his concepts are not wrong, just misplaced till date, coders still learn via hands on and via getting elbow grease on real software work, they may be able to get an opportunity by getting a computer science degree those who relied on paper qualifications almost never begin anywhere current AI models can do basic "coworker" level coding tasks, tasks that are already well known and established, produce bugs, require oversight, but replacing repetition, which is what modern "workers" do. when AI researchers and founders actually starts making real models that can really reason and replace coding there will be huge upheaval and true androids will come we are at infancy but a start
AI is going to flip the script on the requirement being a technical co-founder. A non-technical co founder brings a skillset of knowing how to hustle that you only learn from experience whereas you can hire a technical skillset. There are many projects in the business graveyard, not because they were bad businesses but because they never had a good non-technical co-founder to go to market.
When people said they wanted AI they meant a robot to walk their dog. Why are the only AI we can make the ones that takes peoples jobs 😢? So now people will have no job and still have to walk their dog. 😂
That example of Mickey Mouse and fantasia is so on the nose, isn't the lesson of that story/movie that you should play with things you dont fully understand? And wont agents maybe run amok in ways we cant control? I'm positive about Ai and I'm building with it, but shouldn't we atleast have an honest conversation about what we are doing as Mickey should have told himself, maybe I shouldn't mess with this super powerful spell book since I don't know how the spell truly work? Food for though
![Eminem - Fuel (feat. JID) [Official Lyric Video]](http://i.ytimg.com/vi/t5H_CewqpKA/mqdefault.jpg)
Chapters (Powered by ChapterMe) -
00:00 - Coming up
00:47 - Intro
01:15 - Making an app with Replit
06:19 - Feel the AGI, personal software era
08:07 - Having AI code the way humans do
09:51 - You should still learn to code!
11:42 - The underlying tech
17:19 - The path to AGI
19:41 - What users made with Replit
25:56 - Challenges in resetting the org
33:29 - Future plans
36:12 - Outro
I’ll believe this when YC accepts a non-technical confounding team using these agents to build their products.
Great point!
The only verification 👍
agreed
Only if the industry is correct and the non technical cofounders have brilliant awesome shinny backgrounds lol
But they mentioned in the video, that it's still not a replacement for you taking responsibility of the code, yet. Probably it's only a great tool to get started with a project
As someone whose business's flagship app was built in Replit with help from LLMs, Amjad is way too optimistic. For anything more sophisticated than toy apps, you've got to have software architecting, system and database administering, and lots of software engineering expertise just to pose the right questions to the AI. We're getting there, but it will take at least a couple more years to get where Amjad describes as today.
Couple of years aint long.
@@deeplearningpartnership true that!
휴 다행이다
Im making an app with the help of AI and it's so broken. So this whole talk feels a few years ahead
As a software developer i find the title of this video surreal. IT IS JUST NOT TRUE THAT YOU CAN DEVELOP A SOFTWARE PRODUCT (EVEN A VERY SIMPLE ONE) AND DEPLOY IT WITH JUST AI.
Not yet, but you'll be amazed at the complex systems it can create. I'll share a link before our launch.
Yes you can, I made two Apps on AppStore with 99% ClaudeSonnet 3.5 and Cursor AI.
And much more.
Feels like I have a 10 person team delivering code for me in 10 seconds, that other does in 10 days
It's very true and I've got plenty of acquaintances that have already done it. You just need to be a little patient and a little smart.
As someone who has done product design for the past 30 years, I didn't know how to code, but I did know how to communicate with programmers what exactly I wanted to do. Now I no longer have to rely on other people to realize my ideas. And let me tell you AIs are lot easier to communicate that many humans I had to deal with in the past years…
Replit do that for you already
I used Replit for the first time Yesterday and it Started spinning and got stock in some loop so had to stop it and redo some times o but Eventually it did Create my first LLM API integration and deployed it which made me really happy
I haven't coded in almost 3 decades. Recently, I started using LLMs to write code. It's been fun, I mean REALLY fun because I remembered the days of coding freestyle. However, it has been frustrating too. There is no doubt that the more you know about the language, the faster it goes. I've spent 10+ hours trying to get something to work because of the limitations of LLMs and of course me. I have to prompt the LLMs correctly to reduce the development time and the coding errors. Also, I have to use multiple LLMs to accomplish the task. In my experience, the coding quality is in this order GPT o1 > Claude 3.5 > Gemini > Llama 3.1. If I get stuck in one, I switch to the other to fix the problem. Regardless, this beats learning a language when I just don't have time.
This comment describes my experience exactly. Plus one other point: LLMs are good at writing code and terrible at debugging. Their debugging approach consists of (1) Let's add a ton of logging to find out what is going on, and (2) Let's imagine some strange edge-case scenarios that have nothing to do with reality and code around them with a lot of useless junk. So, even if you don't want to learn a new language, you have to use your development experience to help it find many simple logic bugs. As Harold says, it can still be fun, and I have created extremely useful stuff. Prime example is Apps Scripts for Google Sheets. I am a very experienced spreadsheet user, but I had never created Apps Sheet code before. I am thrilled with what I have been able to do with AI. The scripts are far from being an application, but they can leverage the power of spreadsheets and other tools like Looker Studio to do very useful things.
News flash: Everyone has been able to code for a while now, even before LLMs. There's been a nearly limitless amount of resources to learn coding, many of them for free or at a negligible cost. Not many people do because it takes determination and work to become decent at it. Everyone is looking at LLMs as a free lunch to bypass that work, but code is still being produced by it, and it will still need to be understood if you ever want to create any serious apps.
come back in 1 year and see if u still want to keep your comment or delete it...
@@thebicycleman8062 I don’t delete comments. If you think you can produce millions of lines of code in a production setting without needing to at least verify them before deployment, I’ve got bad news for you. Bugs and unintended behavior can cost you money and customers in the real world.
@@thebicycleman8062 I’ll solve the mystery for you now, I don’t delete comments, especially ones grounded in common sense.
@@thebicycleman8062 I'll solve the mystery for you now. I don't delete comments, especially not ones rooted in common sense.
@@thebicycleman8062 let me solve the mystery for you now, I don't delete comments, especially not ones rooted in common sense.
who is this content for? is YC accepting non-technical founders that make their apps with these tools now?
Yes.
This saves me like 5 second compared to just using Cursor AI that have been out for several months. The drawback is lock-in to replit, compared to Cursor AI that is just a good IDE over Claude Sonnet 3.5 and ChatGPT o1
There's also Continue + Ollama which is 100% free and local
I think the idea is to use Replit as a prototyping tool and if it takes off, you bring in a CTO and move it elsewhere so you don't have "lock in"
If every can code, no one can build a business out of it.
ps: not everyone can code
if everyone is super, nobody is super
-someone from one of the incredibles movies i dont know
@@Peppermynt. Exatcly! That’s Syndrome btw from the first movie
@@Peppermynt. kind of a based quote in this context
Sounds true, but everyone can talk; yet not everyone is a public speaker. Quality, choosing what to build and how to deliver it still will be the differentiators (*even when eventually coding is a fringe role)
@@IntegrandoIA I agree with you. However choosing what to build and how to deliver eventually will not be a topic too. AI will tell you what is the most effective strategy. So how would a business stand out?
Is no one going to talk about the fact that these guys are all avatars?
Unreal what is capable now.
Everyday people say everyone can now code but I am yet to see a single app that a non-developer has built with only LLM. Instead, my workload on upwork keeps climbing. Are these clients not aware they can build their apps with just LLM?😅
This is an excellent roundtable discussion! Note that one of the speakers is talking about the difference between effective AGI and true AGI. It has become quite clear to many of us AI aficionados that effective AGI is just “a few thousand days away”
Not even. I'd say less than 1,000.
I have aged out as a coder, but system architecture and understanding of data is an enduring skill. I would love to build a real application with AI, something with a database, user classes, interactive screens where users can add data, interact with data, and report data. One after another, the "application" examples in these videos are just wrappers around an AI conversation. That is not an application. It's a toy.
what if this toy is actually more efficien nd robust that wht u consider an application in ur traditional sense?
@@thebicycleman8062 Efficiency and robustness are meaningful only after functional requirements are met.
@@SteveKowarsky for me when i said efficiency that includes functional requirments - so efficiency in the sense that it effiiently functions - so yes these new things are not toys anymore, the problem is 99% of A.I critics that come from software dev background of any 'shade' have a personal bias to never want it to work out for obvious conflict of interests, and the problem is this does nothing but just keep their eyes shut down to the extremely rapidly growing oppurtunities and new ways to evolve with the field. How about this Steve, tell me in brief a certain app functinoality or even a tech stack you would consider "a reall application" as opposed to a "toy" and lets have an honest conversation and I will tell you if I have managed to create something similar to it or one of my A.I-Based Coders - Coz I am in a community of a large large group of A.I-Based Developers and you would be surprized what we have managed to build. - I am genuinely interested in knowing what a developer with experience prior to 2023(LLM Widespread Integration) views matters on development in general, I could probably learn a lot from you coz I am definitely aware as an A.I-Based Coder of only 8-month experience that a lot of fundamentals I definitely lack coz I started coding around 8 month ago not even knowing what a code even is, I just thought it was like some hacker matrix thing
AI driven agent-assistant will make engineers life less stressful and also help to spot a bug in a codebase in minutes instead of scrolling through hours
Commenters, the direction is more important than the snapshot. I used it on day 1 and was impressed but it wasn’t fully baked. But it’s a work in progress and this is going to learn faster than most people will. Make your decisions with the assumption that these tools get better every year. They might not out-code a top coder, bit a top coder with it is better than one without. And many businesses don’t have top coders.
Sounds like solutionism. You sure can use these AI Agents to code simple app that will work half the time when you give em the right inputs. Going from a half baked app to a production ready one is where it gets really challenging. And, the bar is so so so high now so I wouldn't say AI agents can build your 'whole' app.
First just a little then all at once
@@ycombinator I agree that current AI tools are useful but often too general-purpose, which limits how deeply they can dive into specific domains-especially when it comes to coding. Even the specialized coding AIs out there still face these limitations. That’s why we’ve taken a different approach, and we’re launching our solution in just 2 weeks.
Our AI agents are designed specifically for backend programming, and we’re already in the testing phase. Unlike most solutions, ours are vertically built, and in certain areas, we orchestrate our AI agents in a more opinionated way. While this can occasionally limit creativity, it ensures that we’re building systems that actually work.
We believe the future of solving every software problem won’t come from one tool but rather the union of specialized systems like ours, each focused on specific areas of development." ı will leave a link here soon
@@ycombinatorlet's cause unemployment for everyone and the rich VCs enjoy a good life
Replit Agent + Cursor + tldraw - Combined into one platform will be super useful
On top of drawing in the UI, it would be nice to be able to take pictures of something sketched on a piece of paper. It's so much easier to use pen and paper as compared to freeform drawing on a screen.
Exactly.
Insights By "YouSum Live"
00:00:30 Personal software revolutionizes app development
00:00:34 Users can build apps in minutes
00:01:30 Live demo showcases mood tracking app
00:08:17 AI assists in coding and debugging
00:09:20 Incremental learning is key for coding
00:20:27 AI tools enhance creativity and productivity
00:22:58 Users create apps faster than ever
00:34:35 Future improvements focus on reliability
00:35:01 Enhanced user interaction through drawing
00:36:01 Advanced users gain more control over code
Insights By "YouSum Live"
I’ll be appearing on this channel in a couple of weeks as a YC founder Insha’Allah
Nice.
What are you building?
dude next to the girl glitched 6:20
A 15 USD/month for Replit Core for using an AGI precursor, it doesn't seem a bad deal when you can play and build a ton. Team AI Aegnt Revolution 2025!
Hi,
please do another podcast about Vertical Ai Agents -
Last one was great!
You could showcase some other SAAS start ups, that are using AI to leverage their business model.
Thank you!
So many people commenting are shockingly critical and pessomistic. It's like seeing the reaction to cameras being added to a flip-mobile-phone back in 1999. Of course it's crap in its infancy.
What they are failing to see is how fast we got here, and where AI could be in mere 5-10 year horizon.
happy they are pushing to still learn to code, as a newbie there's so many clickbait youtube videos saying to stop coding is dead, blah blah.
Should have started earlier this year, exactly same thing in mind
24:43. Roles are swapped, humans are now the coding assistant for AI.
Can you provide link to a meaningful app fully developed by AI?
We are launching our MVP in 2 weeks, and it creates complex apps that actually work. I will edit here and provide a link 🚀🚀🚀(u still need to know how to code)
@@denizturk4307 looking forward to seeing it
Cool.
Nice marketing
But thinking you will replace developers is just silly
Demo is great
Production is different
they didn't say that, though. did you watch the thing?
@@grukoin2789 they are implying it
When are you launching, I'd love to use this for my app ideas😅
You still need to understand how coding works, how algorithms work. Nope not everyone can code.
Not at all
youre wrong.
@@0x0007Nexcept he's right. There's plenty of studies showing that code bases made with AI tend have security issues, a lot of bugs, the majority of commits are just bug fixes for ai generated code and it takes longer to fix as opposed to just writing without AI, leading to a lot of tech debt. If you don't have a technical background, you can't even perceive the flaws in your code
@@KAIZENTECHNOLOGIES Not at all.
Nah, I may not have taken CS classes, but as good as ChatGPT and Claude are at building and debugging, I can accomplish what I need to do.
Human machine symbiosis- bringing human as an extension to Replit I think becomes the most differentiating capability of this tool - Go Amjad. ❤
This is so exciting!! Im HYPED I'm one of those 15 year idea guys...LOL
What's your idea?
If you told Replit the code a Figma clone it would fail horribly 😅 not to mention my app
Everyone can code. What that means at this time is everyone can code simple apps with a minimal outlay of time and knowledge. I now use dozens of personalized apps I generated with AI, often completed before breakfast. Where I might have spent an afternoon making a spreadsheet, I now spend ten to twenty minutes prompting an AI. It's often faster and more reliable that trying to search for an app online that is full of SEO, ads, and unnecessary submit buttons that reload the page for more ads. Real devs can build apps with user logins, persistent state, and lots of features. But normies like me can now build a simple app that uses local storage, import and export buttons, and the exact features they need. There are so many people who sit around wishing someone would build an app that does one specific thing they need, but the apps either don't exist or the features are made for a broad market. Tools like this can serve those individuals.
Can you download the generated code or the database or host it elsewhere?
Yes, from what I remember
The messaging here seems a bit confusing- on the one hand 'anyone' can code, which implies that we will all be interacting on the level of natural language- but there also seems to be a view that learning to code will still be required?
Given the entire history of programming computers is a story of increasing levels of abstraction away from the need to 'speak the computer's language' then isn't the long term outcome of this trend inevitably going to be a programming languge that everyone does already know, which is spoken and written language as used in everyday life?
You can’t say transformers are inefficient learners if you’re just using adam optimizer and back prop
Can I build a native mobile app with replit or is it just PWA.
Neither! PWA is not simple. A.I can outline how to build a pwa very well and concise but it cannot build it for you hence you still have to know how to write code.
I love this dude! Amazing product he is working on
The real goal should be natural language to assembly/machine code.
Great post. But by the way, excel is/was personal software of sorts.
4:04 2023 Mood Logger xD
This is amazing, but strange the app shows all rights 2023?
can someone tell me what Amjad means by Karpathy daigram? at around 31:35
No novel, worthwhile software is being built by LLMs without serious guidance, supervision and instruction on a granular level by a very senior dev.
This is purely moving from 0 to 1.Its like moving from type writers to ms word,
Not everyone can code; you have to know coding to use AI properly for now. Just writing random code is not Software engineering.
his concepts are not wrong, just misplaced
till date, coders still learn via hands on and via getting elbow grease on real software work,
they may be able to get an opportunity by getting a computer science degree
those who relied on paper qualifications almost never begin anywhere
current AI models can do basic "coworker" level coding tasks, tasks that are already well known and established, produce bugs, require oversight, but replacing repetition, which is what
modern "workers" do.
when AI researchers and founders
actually starts making real models that can really reason and replace coding
there will be huge upheaval and true androids will come
we are at infancy but a start
Anyone could code before with effort. Tools have never made it easier.
Replit FTW!
AI is going to flip the script on the requirement being a technical co-founder. A non-technical co founder brings a skillset of knowing how to hustle that you only learn from experience whereas you can hire a technical skillset. There are many projects in the business graveyard, not because they were bad businesses but because they never had a good non-technical co-founder to go to market.
what is baron corbin doing here?
“Mini-Chesky moment” LOL
Please add subtitles
When people said they wanted AI they meant a robot to walk their dog. Why are the only AI we can make the ones that takes peoples jobs 😢? So now people will have no job and still have to walk their dog. 😂
God forbid AI actually solve a real world problem.
That example of Mickey Mouse and fantasia is so on the nose, isn't the lesson of that story/movie that you should play with things you dont fully understand? And wont agents maybe run amok in ways we cant control?
I'm positive about Ai and I'm building with it, but shouldn't we atleast have an honest conversation about what we are doing as Mickey should have told himself, maybe I shouldn't mess with this super powerful spell book since I don't know how the spell truly work? Food for though
For those interested, I made a video using Replit Agent to make a SaaS (no coding experience whatsoever) : ruclips.net/video/7MUKaLqj2oU/видео.html
Who owns the IP?
As long as the AI doesn’t plagiarize non-permissive source code, it belongs to the developer.
Awesome,
guys, what about the nerf of Claude replies?... it s all over reddit
exactly
Oh wow a mood app
Really cool!
pics or didnt happen.
🙏❤️
sure
I have been using Cursor to build my app as a non tech.
These guys are not scaled pilled enough. Too pessimistic about AI progress.
❤❤
❤
lol