AutoGen Studio 2.24 Deep Dive (NO CODING) - Skills, Multi-Agent Workflows, and Real World Case

[@ChrispyKings]

I gotta say you earned my subscribe. I was watching the video you did on just basic set up with Bob & Alice, and I wanted one for AutogenStudio. It was requested in the comments and wow, awesome, you already did it. Thank you

[@anilkumarmamidanna9142]

very detailed , please continue making these autogen video's, need some Realtime examples in related to reading local sql and xml files from local system and perform some tasks.

[@MoBahri]

Thank you for the amazing content, it would be very helpful if you could approximately specify in future videos how much compute power it might take to run X LLM locally.

[@miblish5168]

I need something that starts at a more basic level. What do I do to get the "autogen_studio" prompt? I'm lost after 1:20.

[@kazadori164]

so what do you do if Autogenstudio cant find lmstudio? or is asking for an API key, there is no "local model" options in the model tab of Autogenstudio, openAI, Azure openAI and Gemini are the only options avaibul and I tried with all three, they either asked for an API key or simply cant find lmstudio. is the "local model" you used available for DL somewhere?

[@martfp88]

I just installed autogen Studio with pip install, it installs version 0.1.4, which i believe oa the latest, but group chat is not there anymore :(

[@kushis4ever]

Hello, great tut. I noted you didn't change the model to Llama2 instead of GPT4 when adding the new skill and creating the workflow in the last steps. So, the answers are from GPT4, I would guess.

[@ai-for-devs]

You're correct. The initial part of the video showcases using a local LLM. In the latter part, we shift our emphasis to skill creation without altering the model configuration. However, you're free to integrate both approaches-using different LLMs while crafting skills and workflows to best fit your project needs.

[@kushis4ever]

On another note, do you think we can integrate gemini pro with the autogen as its API is currently free and its abilities are on par with gpt3.5 at least.

[@endash_playlist7710]

great video. is it possible to acess the workflow created in the ui through rest api . i want integrate the workflow with other codes

[@ai-for-devs]

Absolutely! We've enabled similar functionality in 'pure' Autogen using REST APIs and WebSocket connections, as detailed at ai-for-devs.com. This approach should be applicable within Autogen Studio, allowing external access to workflows for integration with other codes. This ensures a seamless and secure user experience. More insights will be shared in upcoming videos, so stay tuned!

[@EngineerAAJ]

why for the skill you used gpt 4 instead of the local model? skills arent avaliable on local models?

[@PubgSpeed-wl8yo]

Is it possible to connect this to the site in the form of a chat? I won’t bring a client to Autogen every time, he shouldn’t see it or be there at all. The chat window should be outside autogen.

[@ai-for-devs]

This is a great idea! We've actually implemented something similar for AutoGen using REST APIs or WebSocket connections, which you can see in action at ai-for-devs.com. I believe we could leverage the same approach within AutoGen Studio. This would ensure that your clients can interact with the chat window without having direct access to AutoGen Studio, enhancing both usability and confidentiality. Stay tuned for video updates in the next few days where we'll dive deeper into this topic and showcase how it can be done!

[@PubgSpeed-wl8yo]

@@ai-for-devs That would be really cool, I'm looking forward to your video about it. Thank you so much!

[@PubgSpeed-wl8yo]

​@@ai-for-devs Microsoft Power Automate Desktop So also look in the direction of this tool, this could be the hands of artificial intelligence

[@PubgSpeed-wl8yo]

@@ai-for-devs And one more small request, make videos for training on youtube, because for example I am not a native English speaker. Videos that are not on KZbin are difficult for me to translate

[@riemannderakhshan1037]

I found it very interesting and novel. However, I could not the source code which you refer through video. It will be more helpful and useful if you share the source code. Are those source code publicly available or only for premium users?

[@ai-for-devs]

When you sign up for our free trial, you'll gain immediate access to the source code and also start receiving weekly exclusive videos, all maintaining the high quality you've found interesting. If you decide it's not for you, canceling is straightforward with just one click.

[@the_original_dreamer]

these are good videos, i would have joined the community if the codes still worked :( guess ill have to wait.

[@francoislepron2301]

I would love to autogen XAML code for describing a UI, because XAML is so boring and not funny at all. Is it possible?

[@TheBestgoku]

The reasons im hesitant to invest time in these things right now is bcz i beleive models will get much smart very soon and all this will be irrelevant. Bcz you can essentially do all this in a single prompt.

[@noobicorn_gamer]

I’ll buy the next phone/console because it’ll be later mentality lol

[@MiguelCayazaya]

like cars, or internet?

[@phily8020-u8x]

Why not have one agent do both create and criticize the joke? Whats the benefit of two in that context?