Activity
Mon
Wed
Fri
Sun
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
What is this?
Less
More

Owned by Rick

Career in Product Development

Private ā€¢ 12 ā€¢ Free

A community for those working in product development

Memberships

Synthesizer

Private ā€¢ 14.5k ā€¢ Free

Skool Community

Public ā€¢ 191.8k ā€¢ Paid

AI Automation Society

Public ā€¢ 3.7k ā€¢ Free

9 contributions to AI Automation Society
Build Smart PDF Automation with LlamaParse
If youā€™re a freelancer working with AI or automation, you know how crucial PDFs are for many businesses. But extracting data from unstructured PDFs? Thatā€™s a painā€”until now. With LlamaParse, you can build smarter automation flows that make handling complex PDFs simple, efficient, and highly valuable to your clients. Why You Should Care ā€¢ Next-Level Parsing: LlamaParse is far superior to the default parse node in N8N, handling even the most complex, unstructured PDF documents. ā€¢ Client-Ready Automation: Working with invoices, contracts, or reports? This guide will help you create a robust PDF parsing flow that your clients will loveā€”and pay for. ā€¢ Win More Projects: Offer tailored, scalable solutions that save your clients hours of manual work and position you as the go-to automation expert. What Youā€™ll Learn 1ļøāƒ£ How to set up LlamaParse in N8N for powerful PDF parsing. 2ļøāƒ£ Extract and structure data from even the messiest PDFs with ease. 3ļøāƒ£ Build a complete automation flow to simplify your clientsā€™ processes, including data organization and storage in Airtable. šŸŽ„ Check out the video tutorial for a step-by-step walkthrough and start building smarter solutions for your clients today!
4
9
New comment 2d ago
Build Smart PDF Automation with LlamaParse
2 likes ā€¢ 4d
@Mark Shcherbakov
0 likes ā€¢ 4d
@Mark Shcherbakov I tested it with markitdown and it gets stripped out also. Llamaparse is the way to go -- but I was open there would be an open source version. All the good solutions seem to be saas though (e.g. firecrawl)
Asynchronous processes
I have a workflow which writes out to a Google Sheet to log messages, but it waits for that node to complete before it sends the reply to the user. I'm wondering is if I can have a workflow that can run nodes asynchronously, so it doesn't need to wait for that one to complete before it moves on to the next. Is that possible?
1
6
New comment 4d ago
0 likes ā€¢ 5d
The default design of n8n is to allow "fan out" -- that means you can have one node connect to many nodes and they will execute async. The inverse is what requires you to use the flow node. Click on the output of the node you want to "fan out" and drag it to your log. It should connect fine
0 likes ā€¢ 4d
Yeah, I just ran into that as well. I gpt'ed the tradeoffs of using n8n vs a framework directly and the lack of parallelism was called out. Your http node usage was the recommended workaround -- glad you pointed it out. -- I'll still leverage n8n since they did an amazing job with credential management and the visually calling is a big plus. I found out you can use the 'Test Workflow' and pass in an array of values for quick "evals". (will not work if you have an aggregate in there though)
first question :) - n8n RAG
Hi :) I am new here, very happy to find this group. What can I do to feed two types of searches in RAG system. 1) We all have it now, the vector search and it works fine. 2) keyword search related to the prompt, so if we use product number like WXZ2432 , it will know it need to share all the chunks it has this unique word. how can we do it?
2
2
New comment 5d ago
0 likes ā€¢ 5d
Metadata is the way to go. During the embedding process, when you are processing the file, there are fields you can extract into metadata. This would still be considered 'one' type of search and a much better route than combining vector results with traditional search results. (for your use case)
RAG with file uploads
I'm trying to build a workflow where an AI agent can use context from a document uploaded in that session by the user. I tried the chat trigger node with upload files enabled, but it requires the document to be sent at every interaction. Has anyone ever tried to build something like this?
2
2
New comment 5d ago
0 likes ā€¢ 5d
if your chat session goes through a vector store, you should only need to upload the document once. The first upload should process the document into the vector store.
Anyone know how to RAG PDFs while preserving links?
The default data extractor node will not consider that information. (or I am doing it wrong) We can create a custom node to extract that data and preserve it with the chunk as metadata, but hoping n8n or the community has already solved this.
1
2
New comment 5d ago
0 likes ā€¢ 5d
I'm using QDrant also (great product). The vector store itself is solid, the problem is with the data extractor. When it processes the PDF to create the embeddings, none of the link information is maintained.
1-9 of 9
Rick Wong
2
12points to level up
@rick-wong-3518
+20 years building as an IC, EM, and Product Manager. I now help SAAS leaders get alignment so their teams can deliver more and build without stress.

Active 2d ago
Joined Dec 18, 2024
powered by