I’ve just come out of a pretty deep rabbit hole setting up my Plex server.

Not my first attempt either.

A mate of mine has a solid setup running on his NAS — Plex, plus Sonarr, Radarr, Prowlarr and Syncthing, all running in containers. I’d tried to copy his setup a few months back using his notes, things I found online, and a bunch of YouTube videos.

It sort of worked. Mostly it didn’t.

I kept running into problems. Versions didn’t line up. The NAS OS looked different to the screenshots. Settings had moved or been renamed. Every guide assumed prior knowledge and setting were exactly the same as the person writing it.

Mine wasn’t. So I left it feeling frustrated.

What was different this time

This time, instead of reading and forums and watching Youtube, I used ChatGPT while I was actually doing the work.

I created a private project and started using screenshots and very specific questions as I went. Permission errors. Folder mappings. “Why does this container see the files but that one doesn’t?”.

It wasn’t perfect — it gets things wrong — but that didn’t really matter.

It felt like having a very capable tech mate sitting next to me while I worked through it, rather than trying to decode advice written for someone else’s setup from five years ago.

The long middle bit

I went down some big days on this. Proper 12–14 hour.

I used the wrong container versions and had to rebuild them.

User permissions were a mess.

Folder mappings were painful — especially when you’re still wrapping your head around how containers actually see storage.

At different points:

• Plex could see files that Radarr couldn’t
• Radarr was downloading things that Plex ignored
• Syncthing was doing exactly what it was told, just not what I wanted

None of it was fast. A lot of it was frustrating. But I was learning the whole time, because I wasn’t just copying steps — I was reasoning through why something wasn’t working.

Where I landed

After hours of breaking things, rebuilding them, and slowly finetuning the configuration of each container and app, I now have a Plex setup I’m genuinely happy with.

More importantly, I understand it.

At the end, I also had ChatGPT help me write base-level documentation for my setup from all our conversations — not polished docs, just clear notes on what I configured and why. I saved it all so when something breaks in a few weeks (because it will), I’m not starting from scratch again.

That alone feels like a win.

The takeaway (without pretending I planned it)

This wasn’t really about Plex.

It was about how different it feels to solve problems with something, instead of jumping between disconnected advice and trying to make it fit your situation.

ChatGPT didn’t magically fix things for me.

It helped me think, test, and keep moving.

And now that this project is done, I’ve caught myself doing the same thing I did at the start:

Looking around for the next rabbit hole.