Hi, thanks for the tutorial and the workflow!! can this be combined with rag for a multi-file research and reporting? what would I have to change look out for?
@derekcheungsa2 ай бұрын
Hi Sebastian, you're welcome. Yes, the easy way to do this would be to hook up a vector store tool to the agents. You can reach-out to me derekcheungsa@gmail.com if you want some specific suggestions based on your use case.
@derekcheungsa2 ай бұрын
I created a sample workflow just now to show case this use case above. It's pretty straightforward.
@giancarlomendes403 ай бұрын
If I want to use the reverse flow: a flow that creates a long document, is it possible with a few changes? If so, can you give some tips?
@derekcheungsa3 ай бұрын
yeah..you just specify the number of words and sections. That is one of the strengths of this automation, you control the length of document to generate
@jlxjalanlinux22403 ай бұрын
it possible to use local ai like ollama , to do agent stuf etc ?
@derekcheungsa3 ай бұрын
Yup, you can. Just switch out the openai node with the ollama one. But, I'm not sure it's worth it. The models are so cheap these days. For example, the google gemini flash 8B is 0.0375/M input tokens, $0.15/M output tokens -- much faster.
@user-wr4yl7tx3w3 ай бұрын
Did you specify which LLM?
@derekcheungsa3 ай бұрын
it's using gpt4o-mini
@ward_jl3 ай бұрын
Very cool! Why do you use different agents per section?
@derekcheungsa3 ай бұрын
Hi Ward, having different agents allows for different prompts to be used for each task. This enrichs the quality of the final output.
@Amplify-AI3 ай бұрын
Amazing!
@Amplify-AI3 ай бұрын
Have signed up to your mailing list but didn’t receive the flow templates. Could you please confirm how we can access these,?
@musumo19083 ай бұрын
Hey nice work. Are the standalone n8n templates available anywhere…as already have self hosted setup thx
@derekcheungsa3 ай бұрын
It's available here: n8n.io/workflows/2457-multi-agent-pdf-to-blog-content-generation/
@MarinasFlorin3 ай бұрын
Where is the workflow located?
@derekcheungsa3 ай бұрын
It's available here: n8n.io/workflows/2457-multi-agent-pdf-to-blog-content-generation/
@coleminephotography20533 ай бұрын
With a 31 page pdf I keep hitting the context window limit at the writers. Suggestions?
@derekcheungsa3 ай бұрын
Yup, you can use Google: Gemini 1.5 Flash-8B, it has 1M context window and is dirt cheap $0.0375/M input tokens $0.15/M output tokens
@derekcheungsa3 ай бұрын
Let me know if you need any help getting that configured. I use openrouter.ai for access to all kinds of models.
@coleminephotography20533 ай бұрын
@@derekcheungsa Got it working with that solution. Thank you.
@user-wr4yl7tx3w3 ай бұрын
Is it free?
@derekcheungsa3 ай бұрын
the template is available for free here: n8n.io/workflows/2457-multi-agent-pdf-to-blog-content-generation/
@lifegritacademy2 ай бұрын
@@derekcheungsa its not for free. You ask for 25USD