It would be cool to have an overview that compares griptape, llama agents, crewai, autogen, maestro and others with their pros and cons. There's so many out there...
This is fantastic way of learning by doing, I love the way you explain the concepts in practical approach. After listening entire session I executed the same code on my Macbook Pro which worked as you explained. Great job Mervin!
I cannot believe that Agents are the future. The cost is too high to let agents decide everything around all possibilities/capabilities for an app. Yes, it is fun to see AI APIs talking to on another, but even Claude Opus, which is definitely some kind of an LLM OS, is very “reduced” in its agentic functionality. It could do more and better, but the cycles that are in need are not used in its full potential.
I think these Agents on individual urls we all already have, because it was not possible for a long time, to load multiple models on the same host of ollama, without avoiding unloading the model, when it changes. now it’s a bit different.
It would be cool to have an overview that compares griptape, llama agents, crewai, autogen, maestro and others with their pros and cons. There's so many out there...
thats a very good idea,all of them are constantly dropping updates ,it would be nice to have like an overview understand of library excels at what ...
thanks. waiting for a more practical demonstration solving a real world problem
Lol, keep waiting😅
Its all about demos and no real value
Thank you
This is fantastic way of learning by doing, I love the way you explain the concepts in practical approach. After listening entire session I executed the same code on my Macbook Pro which worked as you explained. Great job Mervin!
I cannot believe that Agents are the future. The cost is too high to let agents decide everything around all possibilities/capabilities for an app.
Yes, it is fun to see AI APIs talking to on another, but even Claude Opus, which is definitely some kind of an LLM OS, is very “reduced” in its agentic functionality. It could do more and better, but the cycles that are in need are not used in its full potential.
this was awesome. could you share some more complex use cases :)
Wonderful explanation. Thanks for sharing.
A detailed video on your coding environment would be great.... its clear youve been through a couple iterations since I first started following you.
Fantastic 😍
Great vid! idk if i'm not smart enough to deeply understand your example or it could be something more practical..... thanks and carry on!
I think these Agents on individual urls we all already have, because it was not possible for a long time, to load multiple models on the same host of ollama, without avoiding unloading the model, when it changes. now it’s a bit different.
great explanation Mervin
This is amazing!
can it handle multiple user simultanioulsy ?
top top top top 👏👏👏👏👏👏👏👏👏👏
can it run on full local ? with openAI key?
Good job!
When will you cover AutoGroq?
how to run it with local modals,? is there a github repo, looks promising the project.
thanks!d
whats amazing about that? isn’t crewai, autogen and langgraph doing that already?
Will you add this to Praison AI? :D
0:14 ... first mistake ... it's "Loading" not "Logging"
9:08 ... "CallableMessageConsumer" not "Customer"
9:50 "HumanConsumer" Not "Customer"
Thanks for letting me know. Will help me to improve :)
This is not amazing at all. It’s more practical to deploy an agentic workflow to a url not a single agent
Complicate things is not helpful 😢
Teach us something useful
This is useless
I think ITS stupid to call these "agents". They don't act like agentes, they ARE Just instructions on how to deal with information.