I Thought I’d Automated Content Creation. I Hadn’t.

A few months of building AI workflows taught me more about human judgment than automation.

I thought I had fully automated content creation. Turns out I didn’t.

What I Built

A few months ago, I built my first AI workflow. n8n seemed like the best tool out there: you can get Claude Code to create a workflow in JSON and run it on n8n.

I was helping someone with their Instagram, so we tried automating their pipeline: you add a few ideas in an Excel sheet, connect it to n8n, an LLM (I used Gemini API’s free tier) generates various prompts, Flux (an image model via Cloudflare API) creates 2–3 AI images, then the workflow posts directly to Instagram.

I was genuinely fascinated: it worked surprisingly well.

We got ambitious. If Instagram could be automated, why not X and LinkedIn too? We built workflows for both. Content was appearing across platforms without manual intervention.

I thought I’d cracked the code.

But did I?

The Taste Problem

We realized their team was still spending hours refining these outputs.

Sometimes the AI-generated images missed the mark. Sometimes the tone felt off. Not wrong exactly, just… not quite right.

That’s when it clicked: the problem wasn’t speed or accuracy. It was taste.

Creative automation handles repeatable tasks brilliantly. But it breaks on judgment. The last 20% (timing, tone, and how it feels to the audience) needs a human in the loop.

This wasn’t a technical limitation. This was the nature of creative work.

The System Redesign

So we changed the system:

Old: AI generates → Auto-posts
New: AI drafts → Human edits for taste → Publish

We’d moved from “fully automatic” to “semi-automatic”. The results changed everything.

Content quality improved dramatically. The team could focus on strategic decisions instead of mechanical execution. And we were still producing a volume of content a small team couldn’t manage before.

The automation was doing the heavy lifting. Humans were doing the judgment calls.

That’s when I realized: this balance might be the entire point.

What’s Next? The Agent Revolution

The AI landscape is moving fast.

There’s a new tool called OpenClaw (Clawdbot) creating massive buzz. OpenClaw runs on your laptop and phone. Unlike n8n’s fixed workflows, you give it context and it decides what actions to take.

I’m currently running it locally in a sandboxed environment with a local model to limit risk while experimenting.

The question becomes: Do we even need tools like n8n anymore?

Maybe. Maybe not.

I’ve been testing OpenClaw for a week. Still early, but the potential is clear. You have to “train” it, build context daily, teach it your nuances. It’s less about programming workflows and more about teaching a digital teammate.

The Bigger Picture

Here’s the shift I’m seeing:

Agents might handle execution and even parts of decision-making, but human judgment and taste still decide what actually goes out into the world.

This isn’t a limitation. This is the design.

The systems that win won’t be the ones that remove humans entirely. They’ll be the ones that understand where humans add the most value and design for that explicitly.

The early movers in AI automation will have real leverage. Not because they have the best AI, but because they understand how to orchestrate human judgment with AI execution.

That’s why I’m testing everything and sharing what works (and what doesn’t).

The Real Insight

What started as a content automation project became a lesson about the future of creative work.

The creativity isn’t disappearing. It’s moving up the value chain.

The goal isn’t to build systems that work without humans. It’s to build systems that make humans more capable.

Automation handles execution. Taste remains human.

And taste is the only thing that matters.

If you’re curious about the Instagram n8n workflow or how to set up OpenClaw locally (safely and free), feel free to reach out. Happy to share what I’ve learned from both the successes and the failures.

Questions for you: What workflows are you automating? Where are you still spending hours on tasks that feel like they should be automatic? I’d love to learn from your experiences.

#AIAutomation #AgenticAI #WorkflowAutomation #ContentCreation #n8n #OpenClaw