It should not be too hard to edit the py files to point to your own local llm server since they all use the same openai http API syntax. If you can use something like openhermes with it then it will be like crewai which lets you use different llms.
I'm eager to learn free techniques for arming my AI agents with skills, from basic image recognition to specialized tasks like precise PDF formatting (ideal for helping teachers make admin-friendly docs!).
I would love some more depth into the skill / agent creation. I am not sure, but shouldn't it be able to create new skills / agents if needed? Would be nice to use chatgpt but also a local llm, which is directing it in a correct way so we save some money on the long run. But even that could get auto-programmed, right? Would love to get some inspiration how this really functions 😁
I have blindly followed directions on how to set up environments....which is fine when it works - but I had no clue what to do when it didn't. Your patient explanation has not only enabled me to install Autogen ..... but also to understand what I was doing in other projects. This is really appreciated
When you hit a road block just fall back to asking ChatGPT. Give it the context of what you're working with and what you're trying to do, and then give it the direct error message that you received from the terminal. Doing this has increased my coding skills exponentially.
YES! I got this one to work... I spent ~4hrs trying find a method to locate and then install "anaconda-prompt", until AI finally told me you just use Terminal on Mac 🙄 They I spent another 30 minutes trying to successfully add my API key. Again AI finally told me to use "export" instead of "set" and that worked. If anything, I would think "import" but it's working now. Cheers Wes... Looking forward to learning more, in particular how this can be used to help with my workflows. I'm a UX researcher... may AI agent swarm gathering info and loading the findings into a spreadsheet. *hint hint* 😉
As a software developer, this excites and scares me in equal proportions, it seems in not much time, anyone will be able to sit down with these models and create their own software... now. that was the thought on no code, wix, and other technologies and we are still here. I am also starting on AI and currently working on an AI training project, not sure if I'm killing my own future, lol, anyhow, this looks awesome, thanks for sharing.
@@ManPursueExcellence I'm not killing my future, technology advancements always close some opportunities , but open new ones, all you need is to be prepared, adapt and survive
I was reading through their docs. I thought this was hilarious: "Agents keep thanking each other when using gpt-3.5-turbo When using gpt-3.5-turbo you may often encounter agents going into a "gratitude loop", meaning when they complete a task they will begin congratulating and thanking each other in a continuous loop. This is a limitation in the performance of gpt-3.5-turbo, in contrast to gpt-4 wh..."
So you can feel like an actual business owner where employees are being paid to do one thing and spend the rest of the day congratulating each other. Lol. That is awesome. Keeping it human.
Your patient and clear communication is very helpful. Especially for those who have no coding experience. Most tutorials I've found anticipate the viewer has somewhat of an general understanding of little details but pointing out the meanings and terms of various commands helps a great deal to absorb that foundational knowledge, internalize it and run with it. Thanks, your work is well appreciated!
Hey there. Following all your videos. here are few ideas I'd like You to consider making videos on: 1. Autogen vs Chatdev Vs CrewAI; 2. Autogen Autobuild; 3. Obviously the advanced utrorial for Autogen,; 4. Taking Autogen further - into production, without constrains of VSCode, Conda etc... how should it look for totally non tech dudes - let's say Business owners, , who don't need to build swarms of agents but who'd rather use those swarms for their businesses... . It's not for me :), I am builder . Just wondering what your opinion on the topic of how the UI should look like for non technical people
Hi Wes. I absolutely loved your video! Being a teacher myself, I was really impressed by how clearly you explained everything. Right now, I'm diving into the world of AI, and I'm totally fascinated by the concept of agent swarms. I'm actually working on a project to develop a curriculum builder that uses AI to lighten the load for teachers everywhere. It would be incredible if you could think about discussing the idea of building an agent swarm to support educators on your channel. I'd be thrilled to team up and work on this exciting project together! (and yes..I did use GPT to script this lol)
Your explanation has by far been the best keep up the good work. Using your thorough explanation I was able to understand what was going on and troubleshoot my own issue.
Finally a good tutorial on AG studio! Next suggested content: how to integrate AGS in a frontend able to manage input/output like text and files (pdf, images). For example: external frontend > input (text) >> Autogen Studio flow >>> external frontend >>>> output (downloadable pdf with text and images inside). #AutoGenStudioChallenge
Dude, I don't know why the first 10 minutes of this video wasn't in my life 6 months ago LOL ...Good work my friend! Rarely you get somebody that wants to reach back in time a little bit and catch people up if they're not doing this 24/7 like we are.
Create a SWE (think Devin) - Love to see a work flow to create a REST API. The user is a software architect which supplies design documents like swaggers, sequence diagrams and component diagrams. This gets handed off to AutoGen with a Manager which has a Developer, Code Reviewer, Unit test developer, QA engineer which runs code and develops regression tests. All with feed back loops to optimize the code and fix bugs.
For anyone wondering, no, code execution does not seem to work with the Mistral/Mixtral models. The system prompts that Autogen creates are a bit too complicated for these local models to be useful. I’d say wait a few months for some better local models to be released and then try it again.
🎯 Key Takeaways for quick navigation: 00:00 🌐 *Introduction to Autogen Studio's capabilities* - Overview of Autogen Studio's features, - Ability to find research papers, create summaries, and generate images, - Efficiency and time-saving aspects compared to human performance. 02:02 🛠️ *Installation and Setup of Autogen Studio* - Steps for installing Autogen Studio, - Importance of Python and Anaconda in the setup process, - Creation and activation of an Anaconda environment. 07:34 📚 *Understanding Environments and Software Installation* - Concept of environments in Anaconda, - Installation of necessary software and packages, - Setting up and using Python versions compatible with Autogen Studio. 11:32 🖥️ *Exploring Autogen Studio's User Interface* - Introduction to the Autogen Studio UI, - Building skills, agents, and workflows, - Explanation of various components like user proxy, primary assistant, and skills. 18:43 🔧 *Creating Agents and Skills in Autogen Studio* - The process of creating agents and skills, - Utilization of an Autogen Builder GPT for simplifying creation, - Examples of tasks like checking the weather or generating artwork. 22:26 📈 *Future Prospects and Ease of Use* - Future improvements and ease of installation, - The potential of Autogen in various applications, - Encouragement for viewers to engage with Autogen Studio. Made with HARPA AI
I myself am deep diving into Autogenstudio. I am actually using multiple GPT's just to prepare my description of Agent's behaviors ect..I do not shy to run my idea through multiple GPT's to finally end up with AG GPT. Monkey's before they walked went from branch to branch 😊 same here GPT's to GPT's.😊
Of course, I also think that working in this way offers improvements in the quality of the results obtained by the GPTs, however to what extent would they not be sabotaged among the agents in the process?
@@jesusefrenmartindelgado7302 Any time I try out a new GPT there is this simple formula I use here is the basic Idea: "Create five starter conversation that would lead a user to utilize the maximum capabilities of this GPT" The starter conversation give you a good perspective of what that specific GPT can do. They say your never better well served but by one self and in this context is to build your public or private GPT and fine tune them into an Orchestra. Next the beauty is you can use @ to speak to any GPT your private ones and any public one within the same session. Sometime it is impressive and sometime it's not what you wanted Bouya! 🙃 Hey! good luck!
Skills in a work environment. Create documents, create small applications, anything that can help improve work flow creation for normal office workers managing people, vehicles, resources etc
I always use --name 'cos I'm not cool 😢 haha. Great video, thank you! Loved the way you explained Conda, I think that mental model will help new people.
Thanks Wes Roth for the great videos! Somebody was claiming that the new Mamba AI architecture would replace Transformer architecture. Could you create a video on this topic and give us your thoughts? Thanks..
I have three questions. 1) How much would this cost if you were going to write a novel for example? 2) Are there any bat. files made anywhere that can do all of this for you other than downloading python and anaconda? 3) Who else out there is frustrated with how insanely censored Chat GPT is? Thank you again for you very informative and easy to follow videos.
I think the key here is that once good open source models are available, adding them to AutoGen and other similar open source software will make doing this free (or very cheap), un-sensored etc. the price of gpt-4 and the censorship is limiting a lot of uses right now.
9:32 - correction: a locally opened port is accessible from your local network, not just from your computer. for example you could run this on your own pc on port 8081 and then everyone connected to your local LAN or WIFI will also be able to access it. it is however not accessible from outside your local network, unless you explicitly configure port forwarding in your local router.
Hey Wes. Great video. Can you tell me how a person would use this in everyday real world situations? I love it and I’d like to implement it into my workflow. Thanks!!
I would like to see a live example : From start to stop of an agent, incorporation of Dalle 3, downloading images to a floor, and putting its name in EXcel etec and maintaing database
Hi Wes, im a big fan and highly appreciate your hard work and knowledge...I have a proposal though, would you consider having gpt which you'd instruction it as you do on your videos, and share the link of the gpt? on your videos description. I found this quite useful, as sometimes i train a bot on your youtube transcript in order to get better output using your knowledge as an expert :) Cheers
Can you create a video to show how you would create a skill from scratch. Great content and you make is super simple to understand I would also like to see a video on an end to end process including how to use docker. I guess more advanced use cases and how to build great user stories
Thanks for the video Wes. It works on Mac and on Windos 11 with Conda and Python 3.10 (does not work with Conda and Python 3.11 for me). Also installing the autogenstudio only is fine.
This is amazing! I have zero coding experience and was able to install autogen studio, but when I go to the playground to use workflows, I keep getting this: Error occurred while processing message: The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable What did I do wrong?
Did you record part of the video before you installed docker? I am having a much harder time getting it to write things to my desktop or anywhere else on my local machine. It takes a very long time or fails.
Could a social media platform be navigated with AutoGen? Or is there a way to setup a teach mode per say or visual recording / analysis of actions performed?
Is it possible to use this software to create an Ebay listing agent that can use images to look up items and then interface with Ebay to complete listings ?
🎯 Key Takeaways for quick navigation: 00:00 🤖 *AutoGen Studio allows you to build AI agents that can complete tasks efficiently.* 00:55 🛡️ *AutoGen completed a task in 45 seconds that would take a human assistant 30 to 60 minutes.* 02:16 💼 *AutoGen Studio simplifies the process of creating and using AI agents, making it more accessible even to non-programmers.* 04:46 🚀 *Setting up AutoGen Studio involves creating an Anaconda environment, installing required packages, and setting up an OpenAI API key.* 15:58 🎨 *AutoGen Studio can be used to instruct AI agents to perform various tasks, such as generating images based on descriptions.* 19:24 📊 *AutoGen Studio allows for the creation of multiple AI agents in different departments, such as designing, coding, testing, and documenting.* 20:35 💪 *Collaborative efforts of multiple agents working together tend to produce better results than a single agent working in isolation.* 21:30 🚀 *The new AutoGen Studio UI simplifies the process of creating agents, making it easier to define roles and system messages.* 21:43 🛠️ *AutoGen Studio can generate basic code for creating skills, but users need to specify the API and services to be used.* 22:12 🌟 *AutoGen Studio is continuously improving, offering reusable skills and making AI development more accessible, even for non-coders.* Made with HARPA AI
can you explain how the api key works exactly. based on the site it charges you based on use. but for example creating this environment. how expensive can it get?
Everything worked perfectly except the " --n " part. It said its ambiguous and gave me a list of --n's I could use (no lock, no channel, no deps, etc etc) so it made me type out the whole (--name) Either way it worked fine!!!!! TYTYTYTYTYT
btw i just spent 4 hours (lie, it was 5 i think) that create a python "skill" to dl images from the web based on a topic, and preps them (scale to a consistent size, check if webp format is smaller and if so, keeps that one and deletes the original image), to then use for SDXL LoRA training... should've just let my local coding gpt do that
![ลองสุ่มไอโฟน 990 โกงมั้ย? [ โกงมั้ยครับ ep.101 ] | DOM](http://i.ytimg.com/vi/bfsWe0j492k/mqdefault.jpg)
PART 2: The Advanced Tutorial is coming soon! What do YOU want to learn? Comment below!
It should not be too hard to edit the py files to point to your own local llm server since they all use the same openai http API syntax. If you can use something like openhermes with it then it will be like crewai which lets you use different llms.
Awesome stuff! Do you know if there is a way to interface the flows you have built through an API??
I'm eager to learn free techniques for arming my AI agents with skills, from basic image recognition to specialized tasks like precise PDF formatting (ideal for helping teachers make admin-friendly docs!).
I would love some more depth into the skill / agent creation. I am not sure, but shouldn't it be able to create new skills / agents if needed?
Would be nice to use chatgpt but also a local llm, which is directing it in a correct way so we save some money on the long run. But even that could get auto-programmed, right?
Would love to get some inspiration how this really functions 😁
❤❤❤ you are awesome 🚀🚀🚀
I have blindly followed directions on how to set up environments....which is fine when it works - but I had no clue what to do when it didn't. Your patient explanation has not only enabled me to install Autogen ..... but also to understand what I was doing in other projects. This is really appreciated
When you hit a road block just fall back to asking ChatGPT. Give it the context of what you're working with and what you're trying to do, and then give it the direct error message that you received from the terminal. Doing this has increased my coding skills exponentially.
YES! I got this one to work...
I spent ~4hrs trying find a method to locate and then install "anaconda-prompt", until AI finally told me you just use Terminal on Mac 🙄
They I spent another 30 minutes trying to successfully add my API key. Again AI finally told me to use "export" instead of "set" and that worked. If anything, I would think "import" but it's working now.
Cheers Wes... Looking forward to learning more, in particular how this can be used to help with my workflows. I'm a UX researcher... may AI agent swarm gathering info and loading the findings into a spreadsheet. *hint hint* 😉
He literally sais halfway through its set for mac :D
As a software developer, this excites and scares me in equal proportions, it seems in not much time, anyone will be able to sit down with these models and create their own software... now. that was the thought on no code, wix, and other technologies and we are still here. I am also starting on AI and currently working on an AI training project, not sure if I'm killing my own future, lol, anyhow, this looks awesome, thanks for sharing.
You are killing your own future 😂. You’re also building a new one 😉
@@ManPursueExcellence I'm not killing my future, technology advancements always close some opportunities , but open new ones, all you need is to be prepared, adapt and survive
I was reading through their docs. I thought this was hilarious:
"Agents keep thanking each other when using gpt-3.5-turbo
When using gpt-3.5-turbo you may often encounter agents going into a "gratitude loop", meaning when they complete a task they will begin congratulating and thanking each other in a continuous loop. This is a limitation in the performance of gpt-3.5-turbo, in contrast to gpt-4 wh..."
Lmao cute
Reminds me of the managers at a previous employer
So you can feel like an actual business owner where employees are being paid to do one thing and spend the rest of the day congratulating each other. Lol. That is awesome. Keeping it human.
I had to stop my wife from talking to our dog in order to tell her that one.
That’s too funny, bless their digital hearts
Your patient and clear communication is very helpful. Especially for those who have no coding experience. Most tutorials I've found anticipate the viewer has somewhat of an general understanding of little details but pointing out the meanings and terms of various commands helps a great deal to absorb that foundational knowledge, internalize it and run with it. Thanks, your work is well appreciated!
Would love to see a Autogen setup guide running Mistral locally, instead of using OAI API.
Also detailed setup and execution of API(s) .
Hey there. Following all your videos. here are few ideas I'd like You to consider making videos on: 1. Autogen vs Chatdev Vs CrewAI; 2. Autogen Autobuild; 3. Obviously the advanced utrorial for Autogen,; 4. Taking Autogen further - into production, without constrains of VSCode, Conda etc... how should it look for totally non tech dudes - let's say Business owners, , who don't need to build swarms of agents but who'd rather use those swarms for their businesses... . It's not for me :), I am builder . Just wondering what your opinion on the topic of how the UI should look like for non technical people
Not sure if I am ready for that step.... but now I am thinking about it. THANKS!!!
Hi Wes.
I absolutely loved your video! Being a teacher myself, I was really impressed by how clearly you explained everything. Right now, I'm diving into the world of AI, and I'm totally fascinated by the concept of agent swarms. I'm actually working on a project to develop a curriculum builder that uses AI to lighten the load for teachers everywhere. It would be incredible if you could think about discussing the idea of building an agent swarm to support educators on your channel. I'd be thrilled to team up and work on this exciting project together! (and yes..I did use GPT to script this lol)
As a lecturer and innovation manager at an University, I second this proposal ❤
I’m also working on a curriculum builder, but for preschool/pre-k. 👍
I third this ❤
Your explanation has by far been the best keep up the good work. Using your thorough explanation I was able to understand what was going on and troubleshoot my own issue.
Finally a good tutorial on AG studio! Next suggested content: how to integrate AGS in a frontend able to manage input/output like text and files (pdf, images). For example:
external frontend > input (text) >> Autogen Studio flow >>> external frontend >>>> output (downloadable pdf with text and images inside).
#AutoGenStudioChallenge
I wasn't going to attempt install, Wes. But with your encouragement... I'm goin' for it
Dude, I don't know why the first 10 minutes of this video wasn't in my life 6 months ago LOL ...Good work my friend! Rarely you get somebody that wants to reach back in time a little bit and catch people up if they're not doing this 24/7 like we are.
Open source is our biggest hope
❤ I watched over en AutoGen tutorials, but yours is the best in every aspect. 🎉
I almost forgot you code 😅, it’s been a while, I followed you after seeing the first open interpreter video last summer, good to see this!
Create a SWE (think Devin) - Love to see a work flow to create a REST API. The user is a software architect which supplies design documents like swaggers, sequence diagrams and component diagrams. This gets handed off to AutoGen with a Manager which has a Developer, Code Reviewer, Unit test developer, QA engineer which runs code and develops regression tests. All with feed back loops to optimize the code and fix bugs.
Wes, you should DEFINITELY create your own AutoGen Builder. No joke. Just promise to share it with us before you sell it!
This is incredible! I can't wait to start building with it. thank you so much, Wes!
For anyone wondering, no, code execution does not seem to work with the Mistral/Mixtral models. The system prompts that Autogen creates are a bit too complicated for these local models to be useful. I’d say wait a few months for some better local models to be released and then try it again.
ive said it before and ill say it again , so very grateful that you are here with us Wes =] ; stay awesome dood
i love that you are so beginner friendly
Thank you for sharing your knowledge and enthusiasm for learning. I'd like to learn more about AutoGen Agents via your tutorials.
best tutorial on installing and running autogen, excellent
🎯 Key Takeaways for quick navigation:
00:00 🌐 *Introduction to Autogen Studio's capabilities*
- Overview of Autogen Studio's features,
- Ability to find research papers, create summaries, and generate images,
- Efficiency and time-saving aspects compared to human performance.
02:02 🛠️ *Installation and Setup of Autogen Studio*
- Steps for installing Autogen Studio,
- Importance of Python and Anaconda in the setup process,
- Creation and activation of an Anaconda environment.
07:34 📚 *Understanding Environments and Software Installation*
- Concept of environments in Anaconda,
- Installation of necessary software and packages,
- Setting up and using Python versions compatible with Autogen Studio.
11:32 🖥️ *Exploring Autogen Studio's User Interface*
- Introduction to the Autogen Studio UI,
- Building skills, agents, and workflows,
- Explanation of various components like user proxy, primary assistant, and skills.
18:43 🔧 *Creating Agents and Skills in Autogen Studio*
- The process of creating agents and skills,
- Utilization of an Autogen Builder GPT for simplifying creation,
- Examples of tasks like checking the weather or generating artwork.
22:26 📈 *Future Prospects and Ease of Use*
- Future improvements and ease of installation,
- The potential of Autogen in various applications,
- Encouragement for viewers to engage with Autogen Studio.
Made with HARPA AI
Wes explains everything so well.
Need entire series on this...!
thanks
Thank you for walking through the installation. I know it’s simple but very helpful in making these tools more accessible
Good video. Always like the small bits of humour you tuck into the content.
great job Roth, many thanks for this video. looking forward to your future work
I myself am deep diving into Autogenstudio. I am actually using multiple GPT's just to prepare my description of Agent's behaviors ect..I do not shy to run my idea through multiple GPT's to finally end up with AG GPT. Monkey's before they walked went from branch to branch 😊 same here GPT's to GPT's.😊
Of course, I also think that working in this way offers improvements in the quality of the results obtained by the GPTs, however to what extent would they not be sabotaged among the agents in the process?
@@jesusefrenmartindelgado7302 Any time I try out a new GPT there is this simple formula I use here is the basic Idea: "Create five starter conversation that would lead a user to utilize the maximum capabilities of this GPT" The starter conversation give you a good perspective of what that specific GPT can do. They say your never better well served but by one self and in this context is to build your public or private GPT and fine tune them into an Orchestra. Next the beauty is you can use @ to speak to any GPT your private ones and any public one within the same session. Sometime it is impressive and sometime it's not what you wanted Bouya! 🙃 Hey! good luck!
this the most complete tutorial about autogen studio thanks mate ..
Skills in a work environment. Create documents, create small applications, anything that can help improve work flow creation for normal office workers managing people, vehicles, resources etc
I would be very interested in seeing how to determine the workflow between more than 2 agents. Even if it is without the studio ui
Wes, keep up the great work! You're my go to guy for everything AI. Thank you for all you've done and are doing.
Extremely interested in having skills covered over time as it progresses.
this is beyond incredible. wtfffff! thanks a lot for the quick how-to. keep them coming! greetings from buenos aires!
Finally things are getting interesting in this scope. 👍 Great video.
I always use --name 'cos I'm not cool 😢 haha. Great video, thank you! Loved the way you explained Conda, I think that mental model will help new people.
Look guys! It's Not very cool Jordan!
We love you Wesley ! Best AI content on the tube thank you man
Thank you for your tutorial! I was struggling adding the API key on windows.
Thanks Wes Roth for the great videos! Somebody was claiming that the new Mamba AI architecture would replace Transformer architecture. Could you create a video on this topic and give us your thoughts? Thanks..
Dude, you rock. I'm a little more confident using python thanks to your explanation. Subbed like a MoFo.
I have three questions. 1) How much would this cost if you were going to write a novel for example? 2) Are there any bat. files made anywhere that can do all of this for you other than downloading python and anaconda? 3) Who else out there is frustrated with how insanely censored Chat GPT is? Thank you again for you very informative and easy to follow videos.
I think the key here is that once good open source models are available, adding them to AutoGen and other similar open source software will make doing this free (or very cheap), un-sensored etc.
the price of gpt-4 and the censorship is limiting a lot of uses right now.
9:32 - correction: a locally opened port is accessible from your local network, not just from your computer. for example you could run this on your own pc on port 8081 and then everyone connected to your local LAN or WIFI will also be able to access it. it is however not accessible from outside your local network, unless you explicitly configure port forwarding in your local router.
Thank you Wes!!
I really appreciate all your content!
hey Wes, cool video! I was wondering how do you save your agents & workflows? everytime I start the UI instance all my work is gone.
Thank you, Wes. Splendid work.
You should absolutely create the skill creator GPT for autogen studio. Been my biggest hiccup, and i know I’m not the only one.
best explanation by far. thanks
Hey Wes. Great video. Can you tell me how a person would use this in everyday real world situations? I love it and I’d like to implement it into my workflow. Thanks!!
Great Video, i am receiving the following error: ConnectionRefusedError: [WinError 10061]
Hallucination checking agent in the mix - does that help reliability?
I would like to see a live example : From start to stop of an agent, incorporation of Dalle 3, downloading images to a floor, and putting its name in EXcel etec and maintaing database
Great Job WES !
I love your videos :D
great video! thx for your explanations and walk through.
Hi Wes, im a big fan and highly appreciate your hard work and knowledge...I have a proposal though, would you consider having gpt which you'd instruction it as you do on your videos, and share the link of the gpt? on your videos description.
I found this quite useful, as sometimes i train a bot on your youtube transcript in order to get better output using your knowledge as an expert :)
Cheers
Can you create a video to show how you would create a skill from scratch. Great content and you make is super simple to understand
I would also like to see a video on an end to end process including how to use docker. I guess more advanced use cases and how to build great user stories
This video was super helpful!
Wes thank you for another great video!
Thanks for the video Wes. It works on Mac and on Windos 11 with Conda and Python 3.10 (does not work with Conda and Python 3.11 for me). Also installing the autogenstudio only is fine.
Depending on its future competency, I may have to talk to you about this.
This looks so cool!
Did somebody else encounter "Error occurred while processing message: Connection error."? Does anyone know what's the issue there?
same. Did a lot of troubleshooting. no clue yet.
So basically DEI does work 😂😂
Looking forward to playing with autogen thanks for the video homie
Wow didn't know it was this easy to make AI.
That's a follow. My man you content bangs
Amazing content Wes! Thank you so much! What would be the best way to get in contact with you regarding consulting work?
Even though you typed same, it still understood you probably wanted to save the file ^_^
Can we use python venv instead of conda?
How did you get autogen to save directly to the desktop outside of the terminal?
This is amazing! I have zero coding experience and was able to install autogen studio, but when I go to the playground to use workflows, I keep getting this:
Error occurred while processing message: The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable
What did I do wrong?
I got the same issue....any luck?
Phenomenal tutorial, thank you my friend 🙏🏽
Did you record part of the video before you installed docker? I am having a much harder time getting it to write things to my desktop or anywhere else on my local machine. It takes a very long time or fails.
I did my agents and when I went to give it a try, error! Error occurred while processing message: Connection error. So, frustrating!
I think, skill are things that they di, because you develop the function that it does it. And he select that skill when needed
Super useful the costs part
Could a social media platform be navigated with AutoGen? Or is there a way to setup a teach mode per say or visual recording / analysis of actions performed?
15:27 "...an absolutely crucial skill for these AI agents to have" 🤖
Swanky. Thanks Wes.
Think that would be really useful.
That's amazing, can I use a free LLM without a key and no NSFW limitations?
Super helpful. Many thanks
Unfortunatly it doesnt work for me. I keep getting an error message: Error occurred while processing message: Connection error.
Is it possible to use this software to create an Ebay listing agent that can use images to look up items and then interface with Ebay to complete listings ?
Simply incredible!
Brilliant... Cheers! 🏆
🎯 Key Takeaways for quick navigation:
00:00 🤖 *AutoGen Studio allows you to build AI agents that can complete tasks efficiently.*
00:55 🛡️ *AutoGen completed a task in 45 seconds that would take a human assistant 30 to 60 minutes.*
02:16 💼 *AutoGen Studio simplifies the process of creating and using AI agents, making it more accessible even to non-programmers.*
04:46 🚀 *Setting up AutoGen Studio involves creating an Anaconda environment, installing required packages, and setting up an OpenAI API key.*
15:58 🎨 *AutoGen Studio can be used to instruct AI agents to perform various tasks, such as generating images based on descriptions.*
19:24 📊 *AutoGen Studio allows for the creation of multiple AI agents in different departments, such as designing, coding, testing, and documenting.*
20:35 💪 *Collaborative efforts of multiple agents working together tend to produce better results than a single agent working in isolation.*
21:30 🚀 *The new AutoGen Studio UI simplifies the process of creating agents, making it easier to define roles and system messages.*
21:43 🛠️ *AutoGen Studio can generate basic code for creating skills, but users need to specify the API and services to be used.*
22:12 🌟 *AutoGen Studio is continuously improving, offering reusable skills and making AI development more accessible, even for non-coders.*
Made with HARPA AI
Thanks Wes ! You are awesome ❤️
If you can create a Custom GPT that assists on building Autogen workflows, agents, and skills; that would be greatly appreciated!
That was cool!
Working on it!
Great clip
can you explain how the api key works exactly. based on the site it charges you based on use. but for example creating this environment. how expensive can it get?
Everything worked perfectly except the " --n " part. It said its ambiguous and gave me a list of --n's I could use (no lock, no channel, no deps, etc etc) so it made me type out the whole (--name)
Either way it worked fine!!!!! TYTYTYTYTYT
btw i just spent 4 hours (lie, it was 5 i think) that create a python "skill" to dl images from the web based on a topic, and preps them (scale to a consistent size, check if webp format is smaller and if so, keeps that one and deletes the original image), to then use for SDXL LoRA training... should've just let my local coding gpt do that
thank you man!
Looks like fun.
Can we make it do anything? Create websites? Build other code?