N
Hi everyone! I built an after-hours WhatsApp assistant for a study-visa consultancy and wrote up how it works. Sharing it here in case it helps anyone building something similar. What it does Answers student questions only from the consultancy’s own material (services, documents, intakes, offices) Never invents fees, band scores or visa chances, and hands those to a counsellor instead Collects name, service, country and best time to call, one question at a time Replies in the same language as…
N
I am considering how to show traces for agent evaluation, and I noticed that in the UI, we can see the trace in the logs, including tokens, etc. However, that data does not appear to be exposed. Am I missing something? Context: I am using VertexAI models, and while I can make a singular API call with my Service account and use the API response to form the trace and output metadata, I was hoping to keep this classroom-friendly for my students and lean into the AI nodes (Simple LLM or AI Agent…
N
I am struggling to understand the benefit of a pure n8n host my goal is to replace IFTTT (if this then that) (Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.) Information on your n8n setup beforehand questions 5 posts - 5 participants Read full topic
N
I run self-hosted n8n on Railway (Docker, SQLite on a persistent volume mounted at /home/node/.n8n). I previously logged in via GitHub. After what appears to have been a version upgrade during a redeploy, my instance now shows an email/password login instead, and I ended up on a new, empty owner account . My three existing workflows no longer appear. “Forgot my password” returns “n8n isn’t set up to send email right now” because I never configured SMTP. The Railway volume is intact and online…
N
I’m happy to share that we’ll soon be releasing n8n version 3.0.0 (v3 for short), with a focus on reliability, stability, and security. Almost a year ago, I posted about v2, and this release follows a similar pattern: most of the work is under the hood. These are internal changes we needed to make, and they also lay the groundwork for the improvements we’re building next. So why a new major version if there aren’t big new features? Because we follow semantic versioning, we don’t ship breaking…
N
Me ajude, com questões de monitoramento de queda do serviço do n8n, hoje na minha empresa temos o n8n na nossa infra, mas por conta de fluxos pesados e rodando em paralelo, acaba derrubando a ferramenta, hoje temos o datadog pra monitorar logs e erros. Acha que o datadog tem um gatilho que consegue me ajudar com essa notificação ou teria que usar outra ferramenta? 4 posts - 4 participants Read full topic
N
I’m testing the measurement part of the local helper store used alongside my self-hosted n8n publishing project. The case is a metrics collection run that gets a real zero for one field and no value for others. The failure I want to avoid is a later reporting step silently turning “not measured” into “no engagement”. Synthetic input and observed result Today’s focused local test supplied views: 4, posts_count: 1 and responses_from_others: 0, while omitting likes and followers gained. Field…
N
I built a practice workflow to explore how to make batch processing hold up when things go wrong. Sharing it in case it’s useful, and I’d love feedback. What it does Takes a batch of records and validates each one. Sends invalid records down a separate path and logs them Removes duplicates Calls an API with retry handling . Logs successful calls and API errors to Google Sheets. Ends with a run summary (total, invalid, duplicates removed, successes, errors) What I learned Splitting success and…
N
I built a lead capture and qualification workflow in n8n using Apollo, OpenAI, Airtable, Slack, and Gmail. I wanted to share how I structured the workflow, especially the fallback logic and error handling. How the workflow works Webhook – Captures a new lead submitted through a web form. Apollo API – Enriches the lead with company information, including industry, company size, revenue, website, and LinkedIn. AI Scoring – An OpenAI agent analyzes the lead’s message and company data, assigns a…
N
If you move workflows from production to staging with Git source control, the CLI or copy and paste, a copied workflow keeps its Schedule or polling trigger. Staging then runs your production jobs on its own: it spends executions and calls real APIs. This workflow runs nightly on the staging instance and catches that. It lists every published workflow through the n8n API, keeps the ones with a Schedule, Cron, Interval or polling trigger that don’t carry an allow-list tag (default…
N
Hi team, I would appreciate some guidance on private messaging. I joined on October 8 and my profile still shows New User. I have read the Jobs guidelines, but there is no Message button on the author’s profile. The hiring topic is 323862, AI WhatsApp Support Agent, and replies are now closed. The author invites DMs but has not published an email or application form. Could you clarify the normal requirements for messaging, or ask the author to add an accessible application route? I am not…
N
If your Shopify order workflow starts with a Shopify Trigger or webhook, it only runs when the webhook actually arrives. I’ve seen orders silently skipped in three ways: n8n was restarting or updating when Shopify sent the webhook. Shopify retries, but only for a limited time. The execution timed out or errored before the order was logged, and nobody re-ran it. After repeated failed deliveries Shopify removes the webhook subscription completely. From then on the workflow just never runs, with…
N
I run a small AI automation studio in Lévis, Québec. Here’s a workflow now live for a lawn care and snow removal company in Québec City (shared with the owner’s permission). Before: every quote request from the website was retyped into QuickBooks by the owner before he could build the estimate. Now: The website form posts straight to an n8n webhook, with no email inbox in between. The owner gets a notification with a short code and decides whether to create the estimate. On his go, n8n creates…
N
We’re a D2C wellness brand in India selling a digital product with a one-time payment. We get 50–100 customer queries a day on WhatsApp and email, and want an AI agent to handle most of them. What we want built An AI agent on WhatsApp Business that answers common questions (product, access, payments, refunds) in Hindi and English, using our FAQ and policy docs Order and payment lookup via Razorpay so the agent can confirm purchase status Refund requests logged to Google Sheets and routed to our…
N
Hi everyone — I’m part of the team behind Tokatlas, a commercial multi-model API platform with usage-based billing. We’re inviting n8n builders to test how our API fits into a small, concrete workflow and help us identify integration issues. A starting example would be: Sample support messages → model request → category and summary → output validation → spreadsheet The goal is to keep inputs and expected output consistent, then evaluate compatible models on the same task. Alongside answer…
N
This is the tested version I offered in the Tips topic. Set to run every morning at 07:00, it reads yesterday’s orders at a source and a destination, matches them on order_id, and sends one email listing missing, extra and mismatched rows. It sends on clean days too. I wrote “orders and invoices” there; this version compares orders only. What arrives (mixed test day): Subject: Count check 2026-10-22: 2 missing, 2 extra, 2 different Daily count check for 2026-10-22. Source: 10 orders.…
N
Describe the problem/error/question When a Data Table “Get row(s)” node receives several input items and only some or none of them match, the number of output items no longer matches the number of input items. With “Always Output Data” enabled and no matches at all, it emits a single empty item for N inputs. A later node that references a node before the lookup, e.g. $('Split Out').item.json.email, can then no longer resolve the paired item and fails. In my real workflow, I fetch several Jira…
N
getting an error with the Pinecone Vector Store node when using it in “Retrieve Documents (As Vector Store for Chain/Tool)” mode. The index I’m using was created with Pinecone’s new document schema. When the node runs, it fails with this: This index has a document schema, so reads must go through the documents API. Use POST /namespaces/{namespace}/documents/search, /fetch or /list with the X-Pinecone-API-Version header set to 2026-07 or later. The vectors and records read endpoints do not…
N
Disclosure up front: I built the thing in this post and I sell it. I am posting it here because the build details are the useful part, and a few of them cost me weeks. If you want to publish to Pinterest from n8n today there is no core node. What exists on npm is a couple of community nodes that authenticate with a session cookie against Pinterest’s internal API. Those work until Pinterest changes something, and they mean putting your own logged-in session into a workflow. I wanted the official…
N
Describe the problem/error/question What is the error message (if any)? Please share your workflow (Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.) Share the output returned by the last node Information on your n8n setup n8n version: Database (default: SQLite): n8n EXECUTIONS_PROCESS setting (default: own, main): Running n8n via (Docker, npm, n8n cloud, desktop app): Operating system: 2 posts - 2 participants Read…
N
I expected the checker after the sub-workflow call to inspect the result. In the zero-item case, it never ran. Here is the small setup that produced that result: two fictional requests, an embedded child and manual executions in n8n Cloud. No external service was called. The child returned a request ID, a cleaned email value, an ok flag and its input batch size. Three observations: Waiting on: all-items mode and each-item mode both returned the two expected results. The child batch size changed…
N
Questions, bugs and ideas for n8n Labs experiments. They’re early and may change. 1 post - 1 participant Read full topic
N
Quick Answer: n8n is an excellent way to prove that an AI workflow is worth building, but a workflow usually outgrows the prototype once it needs versioned logic, automated tests, strict access control and predictable cost at volume. The practical path is to keep n8n for glue and triggers, move the model-facing logic into code you own, and run both in parallel before switching over. Six teams with public generative AI or integration pages are covered below: Dev Technosys, Vstorm, Softermii,…
N
Hi everyone! I recently built a Procurement PO Checker MVP using n8n, ERPNext API, Excel, and Slack. The problem When a supplier confirmation contains dozens of line items, manually comparing every SKU, quantity, price, and delivery date against the original Purchase Order becomes time-consuming. It gets even more complicated when suppliers confirm quantities in packs or boxes instead of individual units. What the workflow does A supplier confirmation (Excel) is uploaded to Slack. n8n validates…
N
I built a workflow that checks Upwork every 3 hours, scores each new job 0–100 against my profile with Gemini, and only alerts me on strong fits. How it works: Schedule (3h) → Config node (all settings in one Set node) Apify devcake~upwork-jobs-scraper via HTTP Request (run-sync-get-dataset-items) Code node drops jobs with excluded words (“no agencies”, “US only”) Read the sheet once, then remove already-logged jobs by ID Loop one job at a time + 4s Wait to stay under Gemini’s free rate limit…
N
Hello everyone. Over the past few months, I’ve been learning N8N and already built a few workflows for myself. I wanted to expand my skill and turn it into a business to earn some income. However, there’s a problem: I DO NOT KNOW WHERE TO FIND AND TALK TO CLIENTS The idea of going out there and talk to them is terrifying and makes me anxious, but since I am desperate to get out of my comfort zone, I am willing to go out of my comfort zone. I want to target the e-commerce niche, but I’m not…
N
Hey everyone! I built an end-to-end n8n pipeline for DACH/German B2B lead enrichment that chains German directory scraping with an automated § 5 DDG Impressum extractor. What it extracts automatically: Company Name & Legal Form (GmbH, UG, AG, e.K.) Managing Director / Owner (Geschäftsführer / Inhaber) De-obfuscated Direct Email & Phone (+49...) Commercial Register (Handelsregister HRB/HRA) & VAT ID (USt-IdNr.) I open-sourced the full .json template so you can import it directly into n8n via…
N
My n8n Cloud trial was running out, so I set up n8n on my Windows PC with Docker Desktop instead. Here’s the full setup, plus every error and pop-up I ran into, so you don’t get stuck. What you need Docker Desktop (free for personal use) — Windows, AMD64 for most laptops Optional: Claude Code (paid Claude plan) to write the file for you — or just copy the file below 1. Install Docker Desktop Keep “Use WSL 2” ticked. If the installer says “Installation failed. One prerequisite is not fulfilled:…
N
Hi @jamiemadden_n8n, Thanks for moving the Foundations program to Accredible. My Foundations pathway badge and certificate arrived on October 21, 2026. One issue: they were issued under my username “Widget_Julien” instead of my full name. I have since updated my name to “Julien Ansel”, but the credential already issued did not change. Could it be reissued under “Julien Ansel”? Happy to send you my learn.n8n.io email by DM. Credential: Accredible • Certificates, Badges and Blockchain Thanks,…
N
The Jobs category had become noisy, so we are changing how it works to keep it useful without getting in the way. Policy: The category is for hiring posts only. “For hire” and “open to work” posts will be removed. As everywhere on the forum, AI-generated posts are not allowed. Tooling: New topics are closed to replies as soon as they are posted, and a short note at the bottom explains how to follow up. Topics older than 30 days are archived. Applying for a role: Use the email address or…