With how powerful Claude Code is, are we still building automations for clients using n8n, or is everything now being handled directly in Claude Code? What approach are you all taking?

I've been spending time learning how different people use AI and automation in their work, and one thing keeps coming up that I find interesting. A lot of the conversation is about tools, which platform, which model, which integration, but the people who seem happiest with their results tend to talk less about the tools and more about what they decided not to automate. A few ideas I've been figuring out are like: The repetitive stuff isn't always the valuable stuff. It's actually tempting to automate whatever feels annoying, but annoying and high impact are not always the same thing, sometimes the boring task that takes 15 minutes a day matters less than the one that quietly causes errors. Time saved is easier to measure than mistakes avoided. Saving hours is great and easy to point to, but a lot of the real value might be in the things that don't go wrong anymore, which is harder to see and easier to undervalue. Starting small seems to beat starting smart. The most solid setups I've seen didn't start with a big plan, they started with one small thing that worked, then grew from there. I don't think there's one right answer here, a lot of it probably depends on the business and the person. So I'm curious how others think about it, when you decide what to automate, what's your filter? time saved? frequency? something else? Genuinely interested in how people approach this.

There’s a growing belief that artificial intelligence will replace human thinking or creativity. But the reality is far more subtle. AI doesn’t create meaning from nothing—it reflects the quality of the input it receives. It acts more like a mirror than a mind, amplifying your clarity, confusion, depth, or lack of direction. That’s why two people using the same tool can get completely different outcomes: One sees a powerful idea generator The other sees chaotic, unpredictable results The difference is not the AI. It’s the human behind it. In this new era, the real skill is no longer just access to information, but: Asking better questions Defining intent clearly And not outsourcing your thinking to any tool, no matter how advanced it is AI is not the end of human thinking. It’s the beginning of a new phase where thinking itself becomes the real skill. Simple truth: If your question is deep, AI looks intelligent. If your question is unclear, AI looks unreliable.

Hey everyone, my name is Jamie, I wanted to drop an intro post and connect with people building custom automation frameworks. I'm a mechanical engineer by degree, and an automation addict by choice. I’ve spent the last year diving deep into the automation space. I started out messing around with low-code tools like Make and n8n, but I’ve recently hit the limits of what I want to do there. Right now, I'm pivoting toward a more robust, programmatic route - learning advanced webhook management, Cloudflare Workers, and Cloudflare D1 to build lightweight, fast, and highly customized backends. My entry point into this space was actually trying to automate content creation workflows because I wanted a systematic way to handle video production. Once I got a taste of process control here, I became completely hooked on the ability to eliminate repetitive tasks entirely. To be completely transparent, I have severe ADHD. For me, automation isn't just about saving a few minutes; it's a critical tool for managing chaos. If a task is repetitive, my brain drops the ball. Building automated systems is how I create predictable feedback loops to keep my workflows and projects on track. My question for the group: Is anyone else building their automation architecture heavily around the Cloudflare ecosystem (Workers + D1/KV) instead of relying on standard iPaaS platforms? If you’ve gone this route for handling webhooks and data pipelines, what did your learning curve look like, and what are the major engineering gotchas I should look out for early on? Appreciate any insights or architecture advice. Glad to be here.

Hi All! I use Claude Code in VS code and have been heavily using it to test the extent of its capabilities under my AI OS agent. One area I struggle with his getting Claude Code to take action in my browser. Any recommendations on how to set this up (or if you know, is this a known limitation of Claude Code?)