Skip to content

How We’re Using open source and AI to Build FarrierHub: Lessons from the Development Trenches

Building FarrierHub with AI: What Actually Worked (And What Didn’t)

I spent today turning my basic FarrierServices.net site into something that actually works. Not because I had to, but because I wanted to see if this AI-assisted development thing is actually worth the hype.

Turns out, it kind of is. But not in the way you’d think.

The Stack That Actually Works

I’m running Pop_OS! on a Thinkpad P52, using Cursor with Claude, and a bunch of MCP servers that let me actually do stuff without switching between a million different tools.

The magic isn’t in any single tool—it’s in how they work together. WordPress MCP lets me publish directly from my development environment. Discord MCP keeps me connected to the community. Draw.io MCP means I can actually document what I’m building.

What I Actually Did Today

FarrierServices.net was basically a placeholder. Today, I turned it into a real content marketing platform. Here’s how:

WordPress Actually Got Easier

I used to hate WordPress. The block editor was clunky, themes broke everything, and I spent more time fighting the system than building content.

With WordPress MCP, I can manage everything from my development environment. Create categories, publish posts, update content—all without touching the WordPress admin. It’s like having a proper API for WordPress that actually works.

Content That Makes Sense

I set up three main content areas:

This isn’t just marketing fluff—it’s actually useful content that serves different parts of the equestrian community.

Learning From My Mistakes

I made a lot of mistakes today. WordPress blocks kept breaking, themes conflicted with my content, and I had to rebuild the homepage three times.

But here’s the thing: I documented every mistake and turned it into a process. Now I have prompts that actually help me avoid the same problems next time. In fact, this experience gave me an idea: publishing these AI prompts as open source resources.

What Actually Worked

  • Simple beats complex: Every time I tried to get fancy with WordPress blocks, they broke. Simple layouts just work.
  • Mobile-first isn’t optional: If it doesn’t work on mobile, it doesn’t work.
  • Documentation saves time: Writing down what went wrong means I don’t repeat the same mistakes.
  • AI is a multiplier, not a replacement: It makes me faster, but I still need to think and make decisions.

What Still Sucks

  • WordPress block complexity: The block editor is still a pain when you want to do anything beyond basic layouts
  • Theme compatibility: Different themes break your content in different ways
  • Testing across devices: I still need to manually check how things look on different screens
  • Performance optimization: Making things fast is still a manual process

Why This Matters

I’m not building FarrierHub because I think the world needs another app. I’m building it because I’ve been a farrier for years, and I know what actually works in the field.

The AI tools are just making it possible for me to build it faster and better than I could have on my own. But the vision, the understanding of the problem, and the decisions about what to build—that’s all me.

That’s what AI-assisted development actually is: amplifying human expertise, not replacing it.

The Real Takeaway

Today wasn’t about proving AI can build websites. It was about proving that AI can help someone who actually understands the problem build better solutions faster.

The tools are open source. The approach is documented. The mistakes are recorded. Anyone can do this.

That’s the real value: not just building FarrierHub, but showing how to build anything with AI assistance.

Stay tuned for more updates as I actually build this thing. No promises, no hype—just real progress from someone who’s been in the trenches.

Related Resources

Want to learn more about our approach to systems building? Check out these resources:

Ready to build your own solutions? Get in touch or join our Discord to learn more about our approach to AI-assisted development and systems building.

Tags: