This AI Agent can CONTROL your ENTIRE DESKTOP!!!
HTML-код
- Опубликовано: 28 июл 2024
- Introducing the latest innovation in AI technology: Digital Agents that can control your desktop! With ChatGPT, you can now have a virtual assistant that can perform tasks on your computer, just by chatting with it. and this one can operate all of your desktop/web apps.
_______ 👇 Websites👇 _______
github.com/OthersideAI/self-o...
🤝 Discord: / discord
💼 𝗟𝗶𝗻𝗸𝗲𝗱𝗜𝗻: / reda-marzouk-rpa
📸 𝗜𝗻𝘀𝘁𝗮𝗴𝗿𝗮𝗺: / redamarzouk.rpa
🤖 𝗬𝗼𝘂𝗧𝘂𝗯𝗲: www.youtube.com/@redamarzouk/...
www.automation-campus.com/dow...
______ 👇 Content👇 ______
00:00 Understand Digital Agents
01:00 How to install Hyperwrite (Self Operating Computer)
01:32 OpenAI API key
02:15 First Use Case (Desktop Application)
04:21 Second Use Case (Search and Write in Word)
06:35 Main Takeaways
07:48 GPT4V Pricing
Do you think autonomous agents will get a fast upgrade like text to video engines?
missed the oportunity to make it solve the "im not a robot" captcha
That should've been my first thought, I messed up 😂😂
Having my computer managed by an AI I can naturally communicate with is one of my biggest dreams for the short-term future.
I just got this working on my macbook. So this will just run through the terminal?
Can you provide the link the video mentions for info regarding migrating SentinelOne integration in the comments?
Thanks for this video. With a speech-to-text interface it would even be better. And you should try to run it on a GPU for better performance.😉
Yes a siri like assistant that can allow us to create highly customizable automations in a pc is a dream.
Does it work on mbp
I feel you should equally generate content that focus on the free alternatives so people in disadvantaged countries can also benefit...if you see what I mean...
you're right, even though the alternatives are temporary most of the time (because of computing cost), it should be something I talk about.
what if i wanted to use a local llm? like with lm studio. What should i put when asked for an api key?
I tried to use Llava but my machine simply couldn't handle it.
But to be able to run it using a local LLM you need either Ollama or LMStudio (Ollama is easier).
then run the commands:
ollama pull llava
ollama serve
when running hyperwrite use the command:
operate -m llava
it will not ask you for an api key.
Tried it. For some reason it only works with llava 7b and it's definetly too stupid to do anything right. Can't even open the right application or write something in notepad. Wish i could use a bigger model
Hey Reda, I'm having trouble installing it. When I put it into terminal it says: pip : The term 'pip' is not recognized as the name of a cmdlet, function, script file, or operable program.
Hello,
pip is not recognized because most likely the folder Scripts of python is not in the system PATH environment variable.
If you're not familiar on how you can add it, just look up "How to add Python Script folder to Environment Variables".
@@redamarzouk Thank you so much for the fast response, I really appreciate it.
It sucessfully installed but when I put in operate, it said operate : The term 'operate' is not recognized as the name of a cmdlet, function, ect...
@@haroonqureshi9503 after you install self-operating-computer you should be able to have access to the command operate.
Unless you have created a virtual environment in which case you have to access the folder of the virtual environment and then launch the command operate
@@haroonqureshi9503 You probably need to put it in your system's PATH
I wonder how will it work with Groq
well it can only get faster, I have an i7-13620H with 12 cores with a OKish benchmark but using a GPU or even an LPU should make it go so much faster.
Does this get faster as you complete the same tasks?
I haven't notice that, I don't think it has a memory for storing agents. But it will be a nice feature if they have it.
is it for GPT 4 only?
you can use Llava locally if you have a large enough RAM/ of Gemini pro vision if you want another option.
But gpt4v in my experience has performed the best.
Am I the only one who thinks too many people are focusing on software and not hardware. We need to build self-replicating robots(feet and hands) by making the robots be made out of lego blocks that can then be molded with the mold itself also built by the robot.The robots should self-replicate smaller and bigger for whatever you need, and have them build energy collectors, factories, and mining robots.
This is an intermediate step towards that, the physical side is actually a lot harder. Also I'm not sure where you're going with your lego block / mold analogy but I don't follow it.
Still will never pay for AI, free like google sesrch or I'm out.
Yeah I'm with you, I only paid so you guys see the state of AI automation at the moment. but their goal is to reach companies with a good enough offering otherwise it wouldn't make financial sense.
"Free" electricity and GPUs for all?
More AI click bait. GPT4 is untenable for small business $30 for 1M tokens? Look at AIOS it uses gemma and/or mistral just have at least 16GB RAM
In no way you can say it's clickbait, I haven't mentioned the price in the title nor the thumbnail.
I have covered how to use gemma and other models locally and for free in my latest video
ruclips.net/video/C_a52c-1hm8/видео.htmlsi=9XZV_rYXoynn27S1