Where can we speak to you about this as i have some questions regarding this program to check if it suits my requirements ! so please let me know. I love your videos and i am starting to apply but i need relevance to my industry and would like your help. so please let me know where can we contact you or if it is on here no problem. really genuinely interested
Wow thanks man this flow is so much better as the other tutorials! ❤ 1. Project Setup in md files 2. Add documentation in settings 3. The building the Front End 4 Then Connecting it With The APIs 5. Building the Backend 6. Improving the UI 7. Deployment My first flow was A)Setup File B) Vo1 interface C)API integration I was not planning to do web deployment, but maybe I will do Supobase.
I built a crm system in under 6 hours with cursor only.. without v0.. I was sending screenshots of sections of example CRM and I instructed claude that I want mine to look like that and it executed it brilliantly.. and everything without deployment was ready.. this same project took 5 months for a pro dev (JS+PHP) .. I build it with next and it's probably better.. weird times to be a developer
@@wes7bg Indeed man, I just discovered cursor ai and these kind of flows blow up my mind. I've been a software engineer for more than 15 years now and I just quit my job just to focus on this, because this will change the way we develop, the way industry moves forward, and the way we use our tools. I truly believe that we need to spend time on this so that we don't lose our jobs in the future. 6hours crm system seems insane. congratz
I have had a good experience getting the cursor chat window to interactively edit markdown files including product spec docs. So I've been doing this without going off to another external llm service. One trick I found was to tell it that within markdown that it generates for me, to use ~~~ fences for nested code. This is understood by markdown, but not by the cursor chat interface, so it avoids issues with the cursor chat window thinking that the edit has been exited by it entering a nested code block.
One thing is that you need to know what your doing. As an experienced swe all that hes explained are rhe basic of building an app. you have to know what you are going to be building and how the framework work to some degree obviously you might not know all the features of the framework but at least you have a good understanding how it is structured and how you can use it for your application.
Right... you can see he understand software development at a senior swe level. LLMs just speed up his dev time. The idea that anybody can launch SaaS businesses now is comical.
Not true. Just get smarter with prompting. Do research on what methods work best for machine learning snd deep learning. And start throwing those techniques into your prompts. My system: Idea then prompt -> ai expounded prompt -ai refined prompt Then Context(I want an lLM to do this for me) + idea =info on best practices and possible pitfalls Then Ask for a meta set or instructions to achieve refined prompt based on the info gained from context of using an llm to do it for you. Then Ask how the expounded prompt can be optimized to work with the meta instructions. It’s just work but hey it’s quicker thanmemorizing new syntax and how individual languages work.
As we speak it’s cooking for me😂. I told it in interviewing it for a programming job where the test is to take what I’ve given it and get it to fully developed and tested optimized version and told it it’s not a guided test so it’s on its own😂😂😂
@@Keytotransition you still need to understand, even if in a general sense, how the main parts of the code function to do what you describe. This will become increasingly apparent as your app grows in complexity. Yes, I can write a prompt telling the model to design/plan the app like an award winning 180 iq swe, but it still wouldn't function or look, at a nuanced level, how I originally saw it in my head.
@@neilzedd8777 maybe you’re right about that. “need to understand “ I understand now that a program come sun parts called modules and each module needs to be tested and debugged, I understand that setting up a data base appears to be very easy but is. Not. When every module needs specific rules for storage😅. Especially when it’s only saving from what is cached and labeled as successful. But 2 weeks ago I l thought I could just copy files into my venv using file explorer and everything would be okay 👌🏼. I’m saying prompting it to do it for you and having it fail then questioning it is absolutely the most absurdly quickest way to learn so we shouldn’t discourage it by being a stick in the mud.
Could you share total LLM API cost for building this project? By that I mean, how much did your prompts and tokens cost you to build it all (let's omit LLM backend integration from it)
Just want to add that for those of you that are confused or inexperienced about writing detailed spec requirements, you can use the ChatGPT or even Tab in cursor to write them and decide how much detail you want to add.
Wow amazing tips, thanks for ur content 👏. I have a question, why don't you use composer's notepad or custom .cursorrules? Is it for any particular reason?
Thank you for this amazing video. I learned a lot about the workflow, especially the PRD part. But btw I don't think v0 made your UI better. Personally I prefer the original one.
nice job. I've been focusing on ai dev workflow as well, and the process i have so far is really really similar to yours (i developed mine before i saw this vid). I'll share a vid soon solidifying my latest ideas on this.
Really amazing and depressing at the same time. I'm glad I at least learned more complex backend concepts and structures because it seems like frontend devs are cooked at this point.
Amazing thanks you, just a questions when you add a documentation you only add the index url or you must add whole links of the documentation? Thanks you
So i guess we are at the stage that... Like NOTHING we could think up as an idea of design or to code is worth trying to protect as an idea? Just use a tool thats learning off your project because you can do no better anyway....
@@ArtKrishnamurti I agree. AI is just another tool. It doesn’t replace anything. It’s add to the existing stack available to entrepreneurs. All the same business rules / laws / fundamentals still apply.
Is it just me, or the video cuts every 5 minutes. I understand that it is made to shorten video time but I think showing all the parts of creation from beginning to end with the cuted parts will give a lot more insight into what problems you run into and how to approach solving them.
All this is only good for Sunday timepass. Please don't make videos like these. Because my stupid dumb clients watch these and think they can build their MVP with a robust design with just 1 developer doing these kind of things. I am totally pissed off because of these kind of videos where people give false hopes to non tech founders. Full timepass
Comment for more cursor tips you know! If you wanna learn more about Cursor, join my AI builder club: link.agent.rocks/8ni6ADM
Where can we speak to you about this as i have some questions regarding this program to check if it suits my requirements ! so please let me know. I love your videos and i am starting to apply but i need relevance to my industry and would like your help. so please let me know where can we contact you or if it is on here no problem. really genuinely interested
Wow thanks man this flow is so much better as the other tutorials! ❤
1. Project Setup in md files
2. Add documentation in settings
3. The building the Front End
4 Then Connecting it With The APIs
5. Building the Backend
6. Improving the UI
7. Deployment
My first flow was
A)Setup File
B) Vo1 interface
C)API integration
I was not planning to do web deployment, but maybe I will do Supobase.
I built a crm system in under 6 hours with cursor only.. without v0.. I was sending screenshots of sections of example CRM and I instructed claude that I want mine to look like that and it executed it brilliantly.. and everything without deployment was ready.. this same project took 5 months for a pro dev (JS+PHP) .. I build it with next and it's probably better.. weird times to be a developer
@@wes7bg Indeed man, I just discovered cursor ai and these kind of flows blow up my mind. I've been a software engineer for more than 15 years now and I just quit my job just to focus on this, because this will change the way we develop, the way industry moves forward, and the way we use our tools. I truly believe that we need to spend time on this so that we don't lose our jobs in the future. 6hours crm system seems insane. congratz
Possibly the most valuable video for AI-assisted coding. Ever.
This is really a great example of a real structured and professional workflow for the potential role of AI-Assisted Software Developer!
In chat gpt you can click on ctrl+shift+c and it will copy the last response as markdown
I have had a good experience getting the cursor chat window to interactively edit markdown files including product spec docs. So I've been doing this without going off to another external llm service. One trick I found was to tell it that within markdown that it generates for me, to use ~~~ fences for nested code. This is understood by markdown, but not by the cursor chat interface, so it avoids issues with the cursor chat window thinking that the edit has been exited by it entering a nested code block.
Brillian as always! The workflow of how you create the cursor doc is so good
I watched this on X earlier and then could not find it again. Magically it showed up on my UA-cam feed 😁😁 subscribed
Amazing tutorial for me. As a non dev it was a perfect video with exactly the amount of details needed
i have seen a few ai-assisted full stack app development, but this one is really the best structured. thank you!
You dont need to use Claude to get thr markdown version - the openai copy button at the bottom of the response should copy the response content as MD
Thanks for taking time out of your day to always create such methodical and clear instructions. Your amazing ❤ from Germany 🇩🇪
The signal to noise ratio on your videos is amazing
the copy button in openai at the end of the output is in md format
Brilliant tutorial as always. Thank you for your time and sharing with us all.
My favourite AI coding youtuber came through just when I want to build something similar. Thank you!!
One thing is that you need to know what your doing. As an experienced swe all that hes explained are rhe basic of building an app. you have to know what you are going to be building and how the framework work to some degree obviously you might not know all the features of the framework but at least you have a good understanding how it is structured and how you can use it for your application.
Right... you can see he understand software development at a senior swe level. LLMs just speed up his dev time. The idea that anybody can launch SaaS businesses now is comical.
Not true. Just get smarter with prompting. Do research on what methods work best for machine learning snd deep learning. And start throwing those techniques into your prompts.
My system:
Idea then
prompt -> ai expounded prompt -ai refined prompt
Then
Context(I want an lLM to do this for me) + idea =info on best practices and possible pitfalls
Then
Ask for a meta set or instructions to achieve refined prompt based on the info gained from context of using an llm to do it for you.
Then
Ask how the expounded prompt can be optimized to work with the meta instructions.
It’s just work but hey it’s quicker thanmemorizing new syntax and how individual languages work.
As we speak it’s cooking for me😂. I told it in interviewing it for a programming job where the test is to take what I’ve given it and get it to fully developed and tested optimized version and told it it’s not a guided test so it’s on its own😂😂😂
@@Keytotransition you still need to understand, even if in a general sense, how the main parts of the code function to do what you describe. This will become increasingly apparent as your app grows in complexity. Yes, I can write a prompt telling the model to design/plan the app like an award winning 180 iq swe, but it still wouldn't function or look, at a nuanced level, how I originally saw it in my head.
@@neilzedd8777 maybe you’re right about that. “need to understand “ I understand now that a program come sun parts called modules and each module needs to be tested and debugged, I understand that setting up a data base appears to be very easy but is. Not. When every module needs specific rules for storage😅. Especially when it’s only saving from what is cached and labeled as successful. But 2 weeks ago I l thought I could just copy files into my venv using file explorer and everything would be okay 👌🏼.
I’m saying prompting it to do it for you and having it fail then questioning it is absolutely the most absurdly quickest way to learn so we shouldn’t discourage it by being a stick in the mud.
i was waiting for you to drop 😤
Could you share total LLM API cost for building this project? By that I mean, how much did your prompts and tokens cost you to build it all (let's omit LLM backend integration from it)
Would be cheap as, you get an idea really quikcly, but he likely used like
You cannot see the token costs if you're using the pro version. I would say from my experience I spend around 50-100k token per feature.
Started well, but I lost you at 10mins….
This is so good 🙏🏼 thank you for contributing to the community
If you're already preparing detailed product description, you could as well try Pythagora or other tools that create whole project.
This is gold. Thank you!
Good stuff. I always look forward to your next video.
Very helpful! I am going to share this with my community now :)
HOLY SHIT!! THIS IS SO USEFUL thankyou jason
You went further than any other videos, thank you!
How much was the total API cost for this project?
jesus what a banger tutorial thank you!
great example of how to set up a PRD for AI prompting
Just want to add that for those of you that are confused or inexperienced about writing detailed spec requirements, you can use the ChatGPT or even Tab in cursor to write them and decide how much detail you want to add.
can you show us your cursorrules? mine is not getting the code like you
I use a very similar workflow via generated documentation. The v0 example was excellent
Wow amazing tips, thanks for ur content 👏. I have a question, why don't you use composer's notepad or custom .cursorrules? Is it for any particular reason?
This is great, where is the github project?
What if I want to plug ollama instead?
Thanks Bro loved it ♥
Awesome video. Would be nice to have bolt video.
thank you bro superb tutorials
Thank you for this amazing video. I learned a lot about the workflow, especially the PRD part.
But btw I don't think v0 made your UI better. Personally I prefer the original one.
nice job. I've been focusing on ai dev workflow as well, and the process i have so far is really really similar to yours (i developed mine before i saw this vid). I'll share a vid soon solidifying my latest ideas on this.
Really amazing and depressing at the same time. I'm glad I at least learned more complex backend concepts and structures because it seems like frontend devs are cooked at this point.
would you recommend using cursor first then v0 to touch up the UI?
Wow I love it thanks man!
Sorry but how the heck you would know all the packages you are going to use upfront.
Amazing thanks you, just a questions when you add a documentation you only add the index url or you must add whole links of the documentation? Thanks you
The templates you mentioned was very very nice
jason - how come you are able to initialize shadcn without any errors? I keep getting issues that shadcn can't detect a supported framework
Just finish the video is amazing
Very valuable video! Thank you very much!
The era of copy-pasta , the copy pasta context switching is a headache by itself.
shouldn't you be building the components/front end first via v0 then think back on how you want the back end to enable the front end?
What he has explained his good for non developers
Create a .cursorrules file for instructions to cursor
If you do these comments is for the algorithm lol. Great content
Good job! Just referrer code did not pass through...
You're the best guy. Can you do a bolt.new video plssssssssss
So i guess we are at the stage that... Like NOTHING we could think up as an idea of design or to code is worth trying to protect as an idea?
Just use a tool thats learning off your project because you can do no better anyway....
Its either give the idea away for free and let corporations sell it, or see it never happen...
The moat is just switching from development to marketing / distribution.
You still have to stand up a real business, this is just product development. All business fundamentals apply, even more-so now.
@@ArtKrishnamurti I agree. AI is just another tool. It doesn’t replace anything. It’s add to the existing stack available to entrepreneurs. All the same business rules / laws / fundamentals still apply.
Do you happen to have a .cursorrules file that replicates the PRD creation process? Asking for a lazy friend : )
What is this cursor theme?
Is it just me, or the video cuts every 5 minutes. I understand that it is made to shorten video time but I think showing all the parts of creation from beginning to end with the cuted parts will give a lot more insight into what problems you run into and how to approach solving them.
What degrees do you have?
Can you make a tutorial on cursor “notepads”
amazing
Great video, thank you! Cursor has completely changed my coding workflow.
No more junior engineer.
you did stop me from trying to use cursor
just ask the llm’s to make a template for generating a template 😂
This video is gold!! Thanks ❤ I will follow your instructions 🫡❤️
I don’t know, but I know how to know :)
Crazy!
i thought cursor already knew about the file structure, i guess i was wrong
This is amazing video! You are doing very good deed. Just please talk slower :D
Thats interesting am stuck with some basics while using cursor, could you help me out please ?
Whats the issue you are having?
@@Jim-ey3ry Confusion bro..This video has no purpose but to endorse helicone
Good video but why he needs to gargle while talking? Its very annoying
Error Arrows!
Can you slow down? You are not a native English speaker why speak fast?
this
All this is only good for Sunday timepass. Please don't make videos like these. Because my stupid dumb clients watch these and think they can build their MVP with a robust design with just 1 developer doing these kind of things. I am totally pissed off because of these kind of videos where people give false hopes to non tech founders. Full timepass
沙发