Building the workflow is only half the work. The real challenge starts when you try to make the automation reliable. You have to think about things like: → What happens if a node fails? → What if the AI gives an unexpected response? → What if the same request comes twice? → How do you handle edge cases? → Where should a human take over? → How do you make the workflow easy to modify later? At first, I was mainly focused on making workflows work. Now I’m trying to focus more on making them reliable, maintainable, and practical. Still learning, still breaking things, and hopefully getting a little better with every build. 😄 What’s one thing you learned recently that changed how you approach automation?

A few weeks ago I built a fairly complex automation with Next.js on the frontend and Python handling most of the backend logic. Everything looked fine. Tests were passing. Logs looked normal. The automation was doing what it was supposed to do. Then one night around 2 AM, the LLM returned malformed JSON. Normally you'd expect something to fail loudly. This one didn't. Part of the workflow continued, part of it stopped, and the failure wasn't obvious until the client noticed something was wrong. That was the uncomfortable part. The system hadn't really failed. It had failed silently. The client found it before I did. I was honestly lucky they did, because this was a high-value client and it could have turned into a very different conversation. I started looking at how I was monitoring the automation and realised something I hadn't really thought about before: Most of my tooling was very good at telling me what happened. But I needed something that could answer: Did the automation actually produce the outcome it was supposed to produce? So I started building a layer around the automation that doesn't just watch errors. It checks the output against the expected structure, looks for suspicious states, catches things like malformed responses, retries and rate-limit behaviour, records the evidence, and tries to determine whether the workflow should actually be considered successful. The interesting part is that I've found quite a few cases where: HTTP 200 workflow completed LLM said "success" …still didn't mean the business outcome actually happened. I'm about 60% through rebuilding this properly, and I'm beginning to think the difficult part of AI automation isn't getting the workflow to run. It's proving that it ran correctly. Now I'm going down a bit of a rabbit hole with failure reproduction and regression testing too, because fixing a failure once doesn't tell you whether the same thing will happen again three weeks later. Curious how other people are handling this.

Locked in my first client on my Google Maps small business niche project. Trying to nail down a couple of clients in the three niches I picked already got my first one on my first day. started this whole thing after joining Liam’s live stream and ran with the idea. Hope the end ends up working out.

Hello new here. Hoping to learn about AI automation and become a part of the community

A guest just messaged a hotel's WhatsApp number and walked away with a confirmed, priced booking — no human involved. That's the AI booking agent I just shipped for a real hotel use case (Kimti Suites), built 100% in n8n. ✅ **New bookings:** checks real-time room availability (Standard/Deluxe/Suite) against total rooms + existing bookings, calculates total price (nights × rate), confirms details, then books straight into Google Sheets ✅ **Reschedules:** looks up the existing booking, re-checks availability for the new dates, updates the same row ✅ **Cancellations:** confirms with the guest, then updates booking status ✅ **Owner alerts:** auto-emails the hotel owner on every new booking, reschedule, or cancellation ✅ **Filters out noise:** WhatsApp sends delivery/read-status pings on the same webhook as real messages — the workflow silently ignores those instead of erroring out **Stack:** n8n + Google Gemini (AI Agent) + Google Sheets (as the live database) + Gmail + Meta WhatsApp Cloud API This is my first build that goes all the way from a Telegram bot to a full WhatsApp Business API integration, and the WhatsApp setup was genuinely the hardest part — Meta's developer account verification, webhook configuration, and getting a stable public URL for local testing (ngrok's free tier doesn't play well with Meta's webhook validation, ended up moving to n8n Cloud for a permanent URL). 🔑 **Biggest lessons:** - Lock in your sheet structure (columns + format — text vs number) *before* building the tools, not after - Match reschedule/cancel logic on a unique Booking ID from day one, not guest details - A Google Sheets cell that *looks* like a number but needs to match as text will silently fail — format it as Plain Text Open to feedback if anyone's built a similar booking flow or is working through the WhatsApp API setup themselves.