I have problem! I am making auto reply email for my inbox, where AI Agent would check email, and consult with vector database for knowledge and reply within scope that info is available in knowledge database but i would like to approve, reject or give edit before they send email. I tried putting slack node as tool under AI Agent, it didnt consult with me and just go ahead sending email, so i moved the node to outward of ai agent, before sending email, it does send me message with draft email via slack, and i couldn't reply back via free text which i assume because it couldnt go backwards? any ideas? i am open to try out whatsapp or telegram if it works better?

In this video I show a new, faster way to build RAG agents in n8n using the Pinecone Assistant to handle the heavy lifting, no preprocessing pipeline, no manual chunking, no custom embedding flow. You simply drop your documents in and Pinecone takes care of ingestion, splitting, and indexing on the back end. Then we wire it up in n8n to get accurate answers with trustworthy, page-level citations and exact text quotes pulled straight from your knowledge base. If you’re a beginner, this is the easiest path I’ve found to stand up a production-ready RAG agent; if you’re more advanced, it’s a huge time-saver for rapid prototyping. Follow along step-by-step as I connect n8n to Pinecone Assistant, load docs, test grounded responses, and return clean sources you can show to clients or stakeholders, all with a simple, no-code build that you can replicate in minutes.

hey I'm trying to build a whatsApp chatbot that can handle responses both coming from WhatsApp and web chat widget (using freshchat for this) My whatsapp flow is working fine and very well - so basically I build the whatsapp chatbot. now I want to make this workflow or ai agent also handle responses and respond intelligently on a web chat widget for that I'm using webhook so that whenever a user messages something on the web chat widget it receives the messages. So I have two webhooks - one that catches messages from whatsapp and another one that catches messages from web chat widget - both are working fine - I have tested it. however when it comes to merging the output from both with 'merge node' it is only giving me my whatsapp data output and not my web chat widget output. Can anyone tell me why is this happening? Basically my merge node is only passing 1 item instead of passing 2 item.

With big thanks to @Nate Herk's wonderful video tutorials and my assistant ChatGPT 5, over the past two weeks, I built my first automation system in n8n for an NDIS Provider client — and it’s already saving them hours each week. The Problem Their intake process was manual and messy: client info had to be retyped into the CRM, documents were hard to manage, and admins struggled to keep track. The Workflow - Form → CRM: Client fills an intake form, info auto-added to CRM. - Slack Alerts: Extra details go straight to the admin team for quick review. - Docs → OneDrive: Uploaded files stored in the client’s own folder + added to a master CSV list. - Chatbot Bonus: Handles FAQs about services. The Impact - Faster onboarding - Organised documents - Instant team updates - 24/7 client support I’m really proud of this first build — it shows how the right workflow can transform admin work into more time for people.

Hey folks, I built this workflow for an influencer marketing agency that gets 20–30 invoices every week via WhatsApp/Telegram in all sorts of formats — PDFs, Word docs, JPGs, PNGs. Until now, someone was manually extracting all the data into Google Sheets — a super time-consuming process. So I set up this automation: - If the invoice is a PDF, it follows 1st branch. - If it’s a .docx, follows 2nd branch, where it first gets converted, then processed. - If it’s an image (jpg/png), it’s analyzed and extracted in the 3rd branch - If it's an image as a document it goes to 4th branch - Finally, the structured data is pushed into a Google Sheet. The workflow works, but right now I’ve built separate flows for each file type, which makes it long and (probably) more complex than it needs to be. 👉 I’d love feedback from the community on how to optimize/shorten this workflow while still handling multiple file types reliably.