there are many free voices available which can be run locally (but you need a GPU currently for fast inference). But then you will have increased latency
Great video, love the concept of autonomous agents. It seems to be easy to build them. But i hope you do not mind me saying, the humans in the videos need to be more diverse!
Hi! I really enjoyed your videos-this one is very well made as usual. Quick question: what are the avatars that appear at the beginning of this video? By the way, I recently wrote an article titled "SWARMing Conversational AI: Integrating No-Code and Code in Agent-Based Workflows," which you can find online on LI. I’d love to hear your thoughts on my perspective, especially the SWARM emphasis on blending no-code instructions with hardcoded conversational steps. Thanks! Giorgio
Cool overview, thanks for creating! Is the "triage agent" 😎essentially a "router"? Or is there a way to give it management/ hierarchical status or a knowledge base? A use case might be creating article content. If the other agents are out researching and writing content, can the "triage agent" 🧐check the quality of their work and send things back to the other agents to redo if its not high enough quality?
What you're looking for is technically absolutely achievable with SWARM. Yes, the triage agent (or more accurately, an editor/supervisor agent) could work in a back-and-forth process (referred to as hand-off in OpenAI's SWARM parlance) with specialized writer agents in a loop until a certain quality is reached. I can't envision exactly how you’d define the threshold for sufficientarticle quality (that’s not my area of expertise), but perhaps you have insights on this. Some experimentation will be necessary here.
Quick question Chris, Jason Silva has been chatting with AI voice clones of Terrence McKenna and others, really lifelike and the wisdom is in tact...he won't mention the tools used or the source. Could you be kind enough to let us know what is out there that can do that so well? Thank you
Could you use Eleven Labs for audio in this project? The prices are a lot better. Also IBM Watson has a TTS api that works well. I think it's even a bit cheaper than Eleven Labs.
Hi Chris, How do you think this can be useful for building a web crawler that extracts educational curriculum documents? I am trying to build an education app but require the streamlined cleanup of various provinces data into one file, and have been doing this manually before now , don’t want to spend that same amount of time I spent with one curriculum document for the rest of canada provinces figured you may know how to help :) Thanks for making these videos
Please help me understand in what scenarios those agents are better than function calling ? I can understand that when future AI will become so smart that It will be executing multi task assignements that would take hours or days to finish it would be smart to have a couple of (or hundreds) agents and simultanously work on larger project. It seems now it's not very useful.
simplicity! Fucntion calling is all you need, but SWARM facilitates a bitprogramming yourself function calling interfaces, with some compromise on quality maybe.
@@cole1 Thanks, Just wondering why so many people use it, does it allow to clone the code to Github or download locally? I wish you success in your channel bro
I still don't get it how is this useful, I guess simple examples are just not worth doing agents, maybe more complex examples are different story. I think the underlying question is when this will be faster than writing a code doing same thing (insted of prompt you build simple html page with buttons :))? The difference here I guess is that you can give instructions with human language, but so what?
SWARM a part, the fact you can give instructions with human language, that i call '"no code" approach (see my previous comment here) is a huge step forward in computer programming, IMO
THX for Your vids informative & insightful
Thanks for make it easier :-)
Looks pretty good!
How did you combine the female image with the agent??
Did you use some lip since services?
Looks like it is more than likely HeyGen, I could be wrong though
Is there a way to generate the audio locally? It would be nice to generate the avatars locally too. All and all ... Great stuff! Keep baking.
there are many free voices available which can be run locally (but you need a GPU currently for fast inference).
But then you will have increased latency
Great video, love the concept of autonomous agents. It seems to be easy to build them. But i hope you do not mind me saying, the humans in the videos need to be more diverse!
Great job 👏 good video and great used case scenario.
What is the service for making the video? i am working on a educational platform and i would like to create a tutor video.
+1 just curious how did you generate a video??
@@prakruthrs2375 Watch the previous video of Cris about NotebookLM.
seems like heygen
Very good video, gives a lot of idea
Hi!
I really enjoyed your videos-this one is very well made as usual.
Quick question: what are the avatars that appear at the beginning of this video?
By the way, I recently wrote an article titled "SWARMing Conversational AI: Integrating No-Code and Code in Agent-Based Workflows," which you can find online on LI. I’d love to hear your thoughts on my perspective, especially the SWARM emphasis on blending no-code instructions with hardcoded conversational steps.
Thanks!
Giorgio
Cool overview, thanks for creating! Is the "triage agent" 😎essentially a "router"? Or is there a way to give it management/ hierarchical status or a knowledge base? A use case might be creating article content. If the other agents are out researching and writing content, can the "triage agent" 🧐check the quality of their work and send things back to the other agents to redo if its not high enough quality?
What you're looking for is technically absolutely achievable with SWARM. Yes, the triage agent (or more accurately, an editor/supervisor agent) could work in a back-and-forth process (referred to as hand-off in OpenAI's SWARM parlance) with specialized writer agents in a loop until a certain quality is reached. I can't envision exactly how you’d define the threshold for sufficientarticle quality (that’s not my area of expertise), but perhaps you have insights on this. Some experimentation will be necessary here.
If T-Pain was in love with a stripper, then I'm in love with an AI Agent😂
Hmm, wondering making a trading agents. Have you tried do the trading agents your self?
Quick question Chris, Jason Silva has been chatting with AI voice clones of Terrence McKenna and others, really lifelike and the wisdom is in tact...he won't mention the tools used or the source. Could you be kind enough to let us know what is out there that can do that so well? Thank you
Could you use Eleven Labs for audio in this project? The prices are a lot better. Also IBM Watson has a TTS api that works well. I think it's even a bit cheaper than Eleven Labs.
Can it take non-text input?
Hi Chris, How do you think this can be useful for building a web crawler that extracts educational curriculum documents? I am trying to build an education app but require the streamlined cleanup of various provinces data into one file, and have been doing this manually before now , don’t want to spend that same amount of time I spent with one curriculum document for the rest of canada provinces figured you may know how to help :)
Thanks for making these videos
Please email me. My family have an after-school franchisor and could possibly use what you're building.
Great video - thank you. Do you know some tricks how to use o1 models - at the moment officialy it is not possible.
What was that Video thing you've used?
Please help me understand in what scenarios those agents are better than function calling ? I can understand that when future AI will become so smart that It will be executing multi task assignements that would take hours or days to finish it would be smart to have a couple of (or hundreds) agents and simultanously work on larger project. It seems now it's not very useful.
simplicity! Fucntion calling is all you need, but SWARM facilitates a bitprogramming yourself function calling interfaces, with some compromise on quality maybe.
"Why would someone want to have an avatar talking to them while they're on a treadmill?"
Or an avatar talking about wheather or any subject?
How did you generate the talking women? Or are they real? ;-)
HeyGen or Synthesia maybe?
what is the code editor u are using bro?
Cursor
@@cole1 Thanks, Just wondering why so many people use it, does it allow to clone the code to Github or download locally? I wish you success in your channel bro
@@DIY4Profit Cause it has built in features that will write large amounts of code for you with AI. Give it a try! And ty sir
I still don't get it how is this useful, I guess simple examples are just not worth doing agents, maybe more complex examples are different story. I think the underlying question is when this will be faster than writing a code doing same thing (insted of prompt you build simple html page with buttons :))? The difference here I guess is that you can give instructions with human language, but so what?
SWARM a part, the fact you can give instructions with human language, that i call '"no code" approach (see my previous comment here) is a huge step forward in computer programming, IMO
wr,sayx, cna sayx etc infix any nmw s prefx