I use lovable to build a fully sophisticated fullstack web spp, ai is like intermediate developer / engineer, when app grows and gets complicated, it forgets or miss things needed to be done, sometimes it doesnt do what it says, so human intervention is a must especially your app has a few hundred components and sophisticated functionalities 😅
Value is in agents not in ai powered products. as you said clients dont need ai to rent a board, but bord rental companies need to replace the guy who's answering the clients questions all day with a cheap ai.
AI is fantastic but to use it properly and creatively requires a massive skills-set. The idea that less skilled people can level-up their work results with AI is falling flat on the nose earlier or later and represents more a marketing message than reflecting reality.
Nope. It’s 100% a prompt skill issue. AI prompting is the new coding language. You just need to learn the concepts & fundamentals of coding & let the AI do the rest of the work. Prove me wrong 👀
Copilot GPT-4 isn't well enough- put it that way. But, as soon as AI Agents are coded to run things and make humanoids to run things, building with AI is, this way, a different category - not so difficult after AI Agents begin with an LLM like llama 3.1 405 B ?? better, yet, DeepSeek R1 671 B. Am I good at building the API Universal SuperAI System - aftermath aid, please.
great video, i compare working with ai models as working with a very intelligent child, has potential but real work experience and maturity are very lacking
your limited knowledge of how AI works and is properly utilized is apparent here. nobody uses AI chat interfaces to operate AI enabled services and programs, they develop agents in frameworks that leverage the LLMs api only as a natural language based processor while the agents handle the actual work.
Hmm I think I need to rethink how I worded things. All the things you mentioned are the things I’m talking about that are hard. We have people that think it’s easy because they can make it work once in a prompt.
@@GosuCoder I think we should not Generalize things,Lets take Software Industry, it is not just about Coding, There are different aspects such as testing of code. project management, Marketing of software,HR team in the org, other than just coding and managing IT infrastruture, most of the things are getting automated, AI models are very intelligent Adults that are working on real life problems.Here i am just focussing on Software industry,there are other industries too.
@GosuCoder i still think you wildly miss the mark here. You're stuck on using a prompt tonuse AI, and while thats still applicable no matter how you interface with AI, you are thinking about AI as a chat window because most people only have experience with things like ChatGPT for example. Even though you still use prompts in agentic AI implementations it is not quite what you're thinking and alot of it is so very simple that your mind would be blown of you just looked around before you make a video about how you feel AI applies to life based on your limited understanding of what AI is and how it works. For example there are plenty of free open source IDEs and IDE extensions for vscode for example that enable automatic coding, like Roo Code or Aider or Cline for VS Code, Cursor l, Bolt.new/Bolt.diy, Googles Project IDX, Windsurf etc. There are personal assistants that can learn about you and remember what they learn so they become your permanent personal second brain, MSTY for example. The. There are no code/low code workflow automation platforms that use AI to automate almost anything digital you can think of like n8n, Flowise, Langflow etc etc. There are some very nice agent frameworks you can mess with to learn how agentic ai actually works like PhiData. It's important to stop viewing AI as a chat bot or deep fake generator, those are cool abilities but barely even scratch the surface of what's going on in AI right now. I've been self hosing my own AI for quite a while now, and the release of DeepSeek just boosted my abilities through the roof for offline air gapped AI by introducing the world's first open source self hostable reasoning models giving my AI stack the ability to perform critical thought process before giving an answer or building an app or website.
You are absolutely right. It is hard and expensive, not quaranteed to work at all.
VS Code is a product I wouldn’t use without AI anymore ;)
100 percent. I work in a 30k company and we are trying to get it into production.
I use lovable to build a fully sophisticated fullstack web spp, ai is like intermediate developer / engineer, when app grows and gets complicated, it forgets or miss things needed to be done, sometimes it doesnt do what it says, so human intervention is a must especially your app has a few hundred components and sophisticated functionalities 😅
I don't understand the token rate limit. Why don't they just create an AI language that uses far fewer tokens?
Just so damn true.
Value is in agents not in ai powered products. as you said clients dont need ai to rent a board, but bord rental companies need to replace the guy who's answering the clients questions all day with a cheap ai.
🤙🙏
Can use light llm in android phone, which is best to start?
Any suggestions
Can you make video to install usage of it
Think of it as a pseudo-endless stack to pop from.
AI is fantastic but to use it properly and creatively requires a massive skills-set. The idea that less skilled people can level-up their work results with AI is falling flat on the nose earlier or later and represents more a marketing message than reflecting reality.
Nope. It’s 100% a prompt skill issue. AI prompting is the new coding language. You just need to learn the concepts & fundamentals of coding & let the AI do the rest of the work. Prove me wrong 👀
Copilot GPT-4 isn't well enough- put it that way. But, as soon as AI Agents are coded to run things and make humanoids to run things, building with AI is, this way, a different category - not so difficult after AI Agents begin with an LLM like llama 3.1 405 B ?? better, yet, DeepSeek R1 671 B. Am I good at building the API Universal SuperAI System - aftermath aid, please.
Ai is good but the probleme is that the human language is ambiguous thus Ai results will also be
very true! its only a parrot right now
great video, i compare working with ai models as working with a very intelligent child, has potential but real work experience and maturity are very lacking
I really like this comparison!
Alot of hype, click bait, these fake ass guru just want to sell you courses.
what worries me about this is how much people are falling for it :(
Hey, you're absolutely right, but I'm getting good gravy on this train, so keep you damn voice down, eh? 🙂
Skill issue (tldr did not watch the video)
your limited knowledge of how AI works and is properly utilized is apparent here. nobody uses AI chat interfaces to operate AI enabled services and programs, they develop agents in frameworks that leverage the LLMs api only as a natural language based processor while the agents handle the actual work.
Hmm I think I need to rethink how I worded things. All the things you mentioned are the things I’m talking about that are hard. We have people that think it’s easy because they can make it work once in a prompt.
@@GosuCoder I think we should not Generalize things,Lets take Software Industry, it is not just about Coding, There are different aspects such as testing of code. project management, Marketing of software,HR team in the org, other than just coding and managing IT infrastruture, most of the things are getting automated, AI models are very intelligent Adults that are working on real life problems.Here i am just focussing on Software industry,there are other industries too.
@GosuCoder i still think you wildly miss the mark here. You're stuck on using a prompt tonuse AI, and while thats still applicable no matter how you interface with AI, you are thinking about AI as a chat window because most people only have experience with things like ChatGPT for example. Even though you still use prompts in agentic AI implementations it is not quite what you're thinking and alot of it is so very simple that your mind would be blown of you just looked around before you make a video about how you feel AI applies to life based on your limited understanding of what AI is and how it works.
For example there are plenty of free open source IDEs and IDE extensions for vscode for example that enable automatic coding, like Roo Code or Aider or Cline for VS Code, Cursor l, Bolt.new/Bolt.diy, Googles Project IDX, Windsurf etc. There are personal assistants that can learn about you and remember what they learn so they become your permanent personal second brain, MSTY for example. The. There are no code/low code workflow automation platforms that use AI to automate almost anything digital you can think of like n8n, Flowise, Langflow etc etc. There are some very nice agent frameworks you can mess with to learn how agentic ai actually works like PhiData. It's important to stop viewing AI as a chat bot or deep fake generator, those are cool abilities but barely even scratch the surface of what's going on in AI right now. I've been self hosing my own AI for quite a while now, and the release of DeepSeek just boosted my abilities through the roof for offline air gapped AI by introducing the world's first open source self hostable reasoning models giving my AI stack the ability to perform critical thought process before giving an answer or building an app or website.