Activity
Mon
Wed
Fri
Sun
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
What is this?
Less
More
14 contributions to AI Automation Society
Fancy AI agent or boring rules + AI?
Which one do you ship to a client? A) One smart agent that does everything B) Simple rules first, AI only for tagging and drafting, human approves the weird stuff Client won't check on it for a month. Drop A or B and one line on why.
Fancy AI agent or boring rules + AI?
0 likes • 2h
B. First, let rules handle the clear cases, like a new lead with a phone number and a service named, so it gets a reply in under a minute. Second, let AI only tag and draft the reply for the unclear cases, and hold those for a human to approve. Third, set one alert for anything that does not fit a rule, so it does not sit for a month. The pitfall is an agent that tries to decide everything on its own, because one wrong reply to a real customer is hard to undo. What is the first case you would still want a human to approve?
The only n8n roadmap you need to master n8n 🗺️
Struggling to figure out what to learn in n8n and in what order? I made this for you. It's a 10-stage roadmap that takes you from total beginner to power user, with 100 topics in total. Here's what it covers: 1️⃣ Fundamentals: what n8n is, how it works, your first workflow 2️⃣ Core Nodes: HTTP Request, IF, Switch, Merge, Loop and more 3️⃣ Data & Logic: JSON, expressions, aggregate, filter 4️⃣ Integrations: Google, Slack, Notion, Telegram, WhatsApp, Shopify 5️⃣ Workflows: sub-workflows, branching, testing and debugging 6️⃣ Automations: lead capture, reports, data sync, social posting 7️⃣ AI Workflows: OpenAI, Claude, Gemini, AI agents 8️⃣ Self-Hosting: VPS, Docker, security, backups 9️⃣ Best Practices: error handling, naming, credentials 🔟 Power User: custom nodes, n8n API, databases, production scaling 📎 I've attached the PDF with the full roadmap. Save it and tick off each topic as you go.
0 likes • 2h
Good order. I would work through it in three passes. First, build one small workflow end to end, like a form to a sheet, so stages 1 to 3 stick. Second, rebuild that same workflow with error handling and a test run with bad data, which is where stage 9 earns its place. Third, only then add an AI step from stage 7, because the agent is only as good as the data going in. The pitfall I see most is jumping to stage 7 before stage 3, then spending hours fixing JSON and field names. Which stage are people getting stuck on?
How I would automate a new website lead for a service business, step by step
Most service businesses do not lose leads because of traffic. They lose them in the first five minutes, when a new enquiry sits in an inbox and nobody replies. Here is the simple setup I use to fix that: 1. The website form sends the lead to n8n through a webhook. 2. n8n cleans the data, checks for duplicates, and creates the contact in GoHighLevel. 3. The business owner gets an instant notification with the lead's name, number and message. 4. The lead gets an automatic reply within a minute, so they know they were heard. 5. If nobody follows up within a set time, a reminder task is created automatically. The part that usually breaks is step 2. If the fields from the form do not match the CRM fields exactly, leads arrive with missing phone numbers and the follow-up fails. Nothing here needs expensive tools. It needs the boring parts connected properly. How are you handling the first five minutes after a new lead comes in?
New here from Bangladesh - Building for Plumbers / Roofers
Hey everyone, I'm Kabir from Bangladesh. I'm focused on GHL + AI Automation for Plumbers / Roofers. Currently building Missed-Call Text-Back & Speed-to-Lead systems. Starting with the free Classroom and learning Building AI Agents first. Excited to learn from you all. What niche are you working on?
1 like • 2d
Welcome Kabir, Missed-Call Text-Back is the right first system for plumbers and roofers, and I work with service businesses as well. The order I would build it in is: 1. Catch the missed call in GoHighLevel and send the text back within one minute, while the job is still fresh. 2. Ask one simple question in that first text, such as what the job is and when they need it, so the reply tells you if it is urgent. 3. Book or hold a call slot straight from the reply instead of asking them to call back again. 4. If there is no reply in 15 minutes, send one short follow-up, then stop so it does not feel like spam. The pitfall to watch is the phone number. If the number saved in GoHighLevel is not the same as the number the customer called, the text goes to the wrong place and the lead is lost. Are you starting with one trade first, plumbers or roofers, or building the same flow for both at once?
0 likes • 3h
Good call starting with plumbers only. For the follow-up, keep it simple: one short follow-up after 15 minutes, then stop if the job is urgent, because by the next day they will have called someone else. For planned work like a bathroom refit, one gentle check-in the next day is fine, because they are still deciding. The rule I use is never more than two messages without a reply. What kind of plumbing jobs are you targeting first, emergency call-outs or planned work?
My WhatsApp sales agent works in testing but falls apart with real customers. How do I fix it?
I built a WhatsApp AI agent for a local clothing shop in Quetta. Customers text it in Roman Urdu like they would a shopkeeper, and it handles the whole flow: shows products, negotiates the price, collects name, size, quantity and address, saves the order to a sheet, and looks up order status later. What it can do - Replies in Roman Urdu, answers questions about products, shop location and orders - Shows products 3 messages at a time (to save cost), then the next batch if they ask for more - Understands photos/screenshots of a coat the customer found elsewhere and checks if we have the same or a similar one - Handles voice notes (transcribed, then answered) Stack: n8n webhook, code node to drop old/duplicate events, photo/voice to text, AI agent with 3 tools (catalog search, save order, order lookup), a code check on the reply, then the reply is split into text and real images and sent. The problem It works perfectly when I test it, because I know how to order properly. With real customers it breaks. People type casually, don’t know waistcoat terms, and the agent misreads what they want. It sometimes sends wrong products, and after a customer picks one item, it still sends photos of the other products it showed earlier. What I’m unsure about - Is this a system prompt problem, a catalog search problem, or an architecture problem? - Should I add a step that rewrites the customer’s message into a clean query before the agent sees it? - How do you stop the agent from “forgetting” which product the customer picked? I can share the system prompt and a failed chat if that helps. The template is "Free" to grab. Go check it out and let me know what could be changed or improved...
My WhatsApp sales agent works in testing but falls apart with real customers. How do I fix it?
1 like • 2d
Thanks for confirming that. If the photo match is still the weak point, try testing it with 10 real customer screenshots and note which ones it gets wrong. That will show whether the problem is the photo quality, similar looking coats, or the search itself. Once you know that, the confirmation step will catch the rest before an order is saved.
0 likes • 3h
Good luck with it. If the test shows a pattern, like certain colors or angles failing more than others, that tells you whether the problem is the image search or the prompt. Let me know what you find.
1-10 of 14
@ih-tushar-8679
Founder @ Techzious | Websites + AI Agents (n8n, GoHighLevel) for service businesses — more leads, less manual work | techzious.com

Active 1h ago
Joined Mar 28, 2026
Comilla, Bangladesh
Powered by