Theo's Corner
dev / irl / thoughts
← Back
tech

What I'd build next if I had unlimited time

Theo|Jul 2026|~4 min read

A pure brainstorm. No constraints, no "is this realistic," just the projects I'd genuinely love to build if time and money weren't a factor.

A fully self-hosted personal cloud

Not just a NAS with a media stack — a proper personal cloud replacing every SaaS subscription I have. Calendar, notes, photos, contacts, email — all self-hosted, all synced across devices, all under my own domain with a UI I actually like. The pieces mostly exist already. It'd just take a lot of stitching together and polish.

A distributed game server network

Multiple nodes across different regions, automatically routing players to the closest one, seamless failover if a node goes down. Basically a mini CDN but for game servers instead of static files. Genuinely useful, genuinely complex, exactly the kind of thing I'd sink weeks into happily.

A proper offline-first note-taking app

Local-first, syncs when it can, works fully offline, no cloud dependency required. Most note apps get this wrong in one direction or another. Building the exact version I want — fast, minimal, syncing via something I control — has been on the list for a while.

A lot of these ideas exist partly because nothing quite matches what I actually want. Building the thing you wish existed is one of the best reasons to build anything.

A proper home automation system, built not bought

Not a smart plug ecosystem tied to five different apps and five different clouds. One system, self-hosted, controlling everything, with none of it phoning home to a company server unless I explicitly want it to. Home Assistant gets partway there. I'd want to go further.

An actually good self-hosted analytics platform

Something between the simplicity of the stats page on this blog and the depth of proper analytics tools — session tracking, funnel analysis, all privacy-respecting, all self-hosted, no third party ever touching the data. There are options out there but none quite fit what I'd build for myself.

A local AI system, fully offline

Running inference entirely on my own hardware, no API calls out, full control over the model and the data. The tooling for this keeps getting better. Eventually building a genuinely capable local setup, from the ground up, feels like an inevitable project.

Honestly, all of it

Unlimited time means eventually building all of these, one after another, badly at first and then better. That's basically the plan minus the "unlimited time" part — just slower, squeezed in between everything else.