Built an automation that handles insurance claim approvals from start to finish, so your team only spends time on the claims that actually need a human decision. Here's the problem it solves: most claim teams either approve everything manually (slow, expensive) or automate blindly (risky, no oversight). This workflow does neither. It reads every claim, makes a fast decision, and only asks a person to step in when something's genuinely uncertain or risky. What it does: → Reads every new claim automatically the moment it's submitted → Figures out the claim type, how serious it is, and flags anything that looks like fraud → Approves clean, low-risk claims instantly, no waiting → Sends anything risky or unclear straight to your team on Slack, with one-click Approve or Reject buttons → Calculates a suggested payout automatically once a claim is approved → Sends the customer a personalized email update, so they're never left wondering → Keeps a full record of every claim and decision in Google Sheets, so you always have an audit trail Who this is for: insurance teams, claims processors, or any business drowning in repetitive claim reviews who want to speed up the easy cases without losing control over the risky ones. Try it out. Free Workflow link in the comment.

Built an n8n community node for Jev (TypeSafe AI's new decision model) Unlike a plain API wrapper, this node routes your workflow automatically based on Jev's answer and how confident it was. What it does: → Send any text/data + your own questions (Choice, Score, or Noul/Yes-No) → Get back Jev's answer with a confidence score attached → Node auto-creates output branches for each possible answer (like a Switch node, but AI-driven) → Low-confidence answers route to a separate "Needs Review" branch instead of auto-processing Bonus mode — Calibration Check: Feed it your own labeled historical data (tickets, leads, whatever you already know the right answer for), and it tells you Jev's real accuracy at different confidence levels — so you set your threshold based on data, not a guess. Built it for classification/routing use cases like ticket triage, lead scoring, and content filtering — anywhere you're currently burning a full LLM call just to get a one-word answer. How to install: 1. In n8n, go to Settings → Community Nodes → Install 2. Enter the exact package name: n8n-nodes-jev-router 3. Hit Install 4. Search "Jev Router" in the node panel, add your TypeSafe API key as a credential, and you're set Would love feedback if anyone tries it out.

Built a workflow that onboards hourly/frontline staff (retail, warehouse, restaurant) without any manual paperwork checking. Here's what it does: 📲 New hire messages a Telegram bot to start onboarding 📄 They send a photo or PDF of their ID — AI (DeepSeek) reads it automatically and pulls out name, ID number, expiry date, and checks if it's genuine ✅ Once verified, they get the company policy and reply AGREE to confirm 🔔 Manager gets notified the second onboarding is complete ⏰ Anyone stuck for 48+ hours gets an automatic nudge (and so does their manager) 🗂️ Everything lives in Notion as a live dashboard — no login needed, managers just check the page No HR software, no manual ID verification, no chasing people down. Just a chat bot + AI + a Notion tracker doing the whole thing. Built this for businesses that can't justify enterprise HR tool pricing but still need real onboarding compliance. Workflow link in the comment

Wrapped up the n8n Professional Certificate program this week. I already build in n8n regularly, but I went through it anyway to revisit the fundamentals properly. Glad I did, because a few things I'd been doing on autopilot finally clicked. It's 3 courses, all hands-on with real exercises (not just watching videos): 🔘 Essentials: Your First Workflows - triggers, core nodes, workflow logic 🔘 Integrations: APIs & Connected Workflows - connecting third-party apps and handling real API scenarios 🔘 In Practice: AI, Testing & Best Practices - adding AI to pipelines and testing them properly before shipping If you're just starting out, this is a solid path. If you've been building for a while, it's still worth it for the testing and best practices part. Big thanks to n8n Academy for putting together such a well-structured course.

Most real estate teams still handle leads the same way. Someone fills a form, sends a WhatsApp message, or clicks a Facebook ad, and then waits. Meanwhile the lead sits in an inbox until an agent has time to look at it. I built an automation that removes that wait entirely. Here's what happens the moment a lead comes in: The system checks if this person has already reached out before, so nobody gets a duplicate follow up. AI reads their message and figures out what they actually want. Buying, renting, or just browsing. How serious they are. What their budget looks like. It then picks whichever agent has waited longest without a new lead, so leads get shared fairly instead of going to whoever's fastest to check their phone. The lead gets saved with all of that information attached, the agent gets an email with the details, and the lead gets a message back with a link to book a call. All of this happens in seconds, with nobody touching a spreadsheet. Link to the free workflow is in the comments