Anyone have any experience/advice
Anyone have any experience/advice building agentic workflows? Short-term question is: use a framework or build the workflow myself?
9 Replies
I'm starting my journey building agentic workflows, starting with a translator tool.
I have seen some criticism of LangChain that it has a black box feel and that the industry moves fast enough that you shouldn't rely a framework
https://blog.langchain.dev/what-is-an-agent/
By agentic, I mean AI agents
Matt Williams
YouTube
Have You Picked the Wrong AI Agent Framework?
Are you diving into the world of agent-based AI workflows and finding it more complicated than it needs to be? You’ve probably chosen the wrong approach, burdened with verbose, time-consuming frameworks. In this video, I’ll show you a far easier method to build your AI workflow, cutting through the unnecessary boilerplate and complexity.
Using ...
He is from https://ollama.com/
I would look into xstate and ollama and watch some of his vids. He is fun
Ollama
Get up and running with large language models.
I would think convex llama farm would be the most helpful example?
Thanks @ampp - Interesting, llama farm must be taking the output of one LLM and passing it to another. thanks!
Yeah, i cant recall the exact setup but i think if every agent was identified by a type then it would be rather easy. I might try to do something like llama farm but have it communicate the embeddings for some highly custom stuff. 😅
We aren't a ai project but we need every user to connect their own agent farm if they desire
I feel like what I'm trying to do is much simpler than Llama Farm is. I want my end user to be able to translate a long text, and then edit parts of it (by making another AI prompt), and keep iterating until they are satisfied
Yeah, that sounds like you can get away with more of a standard setup with like streaming responses.