Welcome to ScaleAtHome
I’ve always wanted to live in the future
I got my first taste of that future booting up SimCity on an old DOS PC in our childhood basement. “What do you mean I can build a whole city? This is magic!”

Over the years that magic slips away bit by bit into the routine. Every so often we get a jolt. In early 2026, I opened my first Claude Code session and found myself disoriented. It certainly feels like we’re exploring fresh territory with a whole set of magical new tools.
But a new normal creates new challenges: subscriptions stacking up, data living on someone else’s servers, and the nagging feeling that I was renting my own productivity. I wanted local control. I wanted privacy. And honestly? Building this stuff yourself is just fun and interesting in a way that clicking “subscribe” never will be.
“We are stuck with technology when what we really want is just stuff that works.” — Douglas Adams
That quote nails it. These new tools pushed me toward a simple conclusion: we’re on a new track toward updated personal computing, and I want to own as much of it as I can. Local AI, local infrastructure, local data.
This site documents that journey and shares the hard lessons learned along the way.
The site
Not a review site. I’m not scoring products out of 10 or ranking the “7 best mini-PCs of 2026.” What I do is document my experience with the gear and software I actually run — what worked, what didn’t, and what I’d do differently.
Think of it as field notes from someone building and maintaining local infrastructure every day:
- DIY Data Collection - Weather, around the home, personal data to power life
- Refurbished hardware — enterprise gear at consumer prices.
- Local AI hardware — mini-PCs and GPUs for running LLMs locally with Ollama, llama.cpp, LM Studio.
- Storage — DIY builds with real-world reliability data.
- Networking — Tailscale and the other infrastructure that ties a homelab together.
- Self-hosted software — Linux, Docker, monitoring stacks, and the tools that replace cloud SaaS.
- Developer & AI tooling — coding against local infrastructure and running AI agents on your own hardware.
The bias
I have one, and I’m upfront about it: value-tier, refurbished, and self-hosted over mainstream cloud. If a $200 refurb OptiPlex does 80% of what a $600 new mini-PC does for your use case, I’ll tell you that.
The gear
Here’s a taste — one of the refurb SFF desktops that powers the setup:
HP
HP EliteDesk 800 G4 SFF
~$250
- cpu
- Intel Core i5-8500 (6C/6T)
- tdp
- 65W
- gpu
- Intel UHD Graphics 630
- ram
- 8GB DDR4 (upgradeable to 64GB)
- storage
- 256GB M.2 NVMe
- form-factor
- Small Form Factor
The future
Subscribe to the site RSS or check back regularly for new entries in this story.