LangGraph Engineer
HTML-код
- Опубликовано: 11 сен 2024
- Try out the deployed version: smith.langchai...
This is an alpha version of an agent that can help bootstrap LangGraph applications. It will focus on creating the correct nodes and edges, but will not attempt to write the logic to fill in the nodes and edges - rather will leave that for you.
LangGraph: github.com/lan...
The agent consists of a few steps:
1. Converse with the user to gather all requirements
2. Write a draft
3. Run programatic checks against the generated draft (right now just checking that the response has the right format). If it fails, then go back to step 2. If it passes, then continue to step 4.
4. Run an LLM critique against the generated draft. If it fails, go back to step 2. If it passes, the continue to the end.
So freakin kewl, I could almost cry. Tooling like this is easy to imagine as the next evolutionary step. And you're creating it!
Please ensure the documentation is rich as features are developed. Not being able to use tools for lack of documentation is as painful as not having the tools at all. Thanks!
Loving these new Langgraph apps. I’ve had langgraph studio up and running for a few days and it’s been a nice addition to my ide.
I just tried it out. I pasted in my LangChain tool names with an example of how to initialize them and described the logical flow I wanted to achieve. It quickly generated a workflow with three LangChain agents. Although there’s a limit to the output length, when I pasted the last line and asked it to continue, it seamlessly provided the rest of the flow. It worked perfectly right out of the box and offered a comprehensive solution with a few conditions. Really impressive!
Still awaiting for you to bring langgraph studio for windows. Currently, it only available for mac users as dmg file 😫
There are several reports of users willing to pretend Mac ownership having a really good chance to pretend success installing the dmg file.
I have been developing for over 30 years. This video hits it right on the head with what I’m witnessing. In this age “it’s going to be too hard” is a metric 😮
In the kingdom of AI Agents you are the king 👑
So hyped for the LangGraph customer support ticket agent video coming soon.
Haven’t seen watched it yet… but I’m anxious to do it first thing when I get back home. If the title has anything to do with the content this it exactly what I wanted to work on myself ❤
I have a feeling with the new openAI model this will work well. With the code most of it is just fill in the blanks so you could tell the LLM exactly what you are expecting and it would just fill in the blanks. One thing that would be sorta cool is having drag and drop nodes where you just filled in the variables. You could even have custom nodes that people could save. Think sorta like what Blender is. There is huge potential here. One thing I've found so far is unless you knew code for the most part people needed to know how to code in order to really utilize this stuff. There is no reason it shouldnt just all be drag and drop. If a few of the open source development teams got together to build a team to work on a LangGraph front end you could move pretty quickly. Langchain is dealing more with the agent system while LlamaIndex is dealing with the document side of things. DSPy is dealing with the prompting. You could even have Ollama involved to run models locally but it would also be great for a help system. If all the documentation is very well structured then it would be easy for a small LLM to guide users along if they have any issues.
This is awesome, thanks.
Nice! When Windows support for Studio?
or WSL2 :)
Thanks for the info and all of your work. I have a quick question PLEASE answer: I can get this all to work locally, but I am unsure of how to get input and Human in the Loop input in the context of Langgraph Cloud. Any pointers? Thanks again.
I tested it and it's just a magic. it speeds up the workflow very much, please add javascript support
Where can I find the python code of the app? Also, Harry, you are a master of Gen AI. Thank you for the great job you and the team are doing!
This is stunning work!! Thanks for sharing!
does it persist the state? is it possible to horizontally scale ?
Awesome work, will be using this
Thank you once again! Huge fan 🙏
Awesome! 🎉 we need more of that, think about business logic, less about code…
How can I use pdf to rag responses here? Like can I load up pdfs or docx etc as data to draw on to generate responses?
This is just awesome!
i'm testing this like a crazy 😁
love your work 😍
You are a hero
awesome!
great work
Very awesome!
how can i run it on command line?
Timing for TS?
You guys are fucking amazing!
nice.
bro isn't it about time you start wearing a leather jacket 😋
let another agent fill in the logic!
Gonna need linux support before I can use it.
dude go fix the docs first. fuckin hell
We want Windows Support Please