I’m building a new home technology stack with the idea that I use as much open source tooling as possible to self host many of the services I currently pay for from Google, Apple, Microsoft, and others.
This environment still has some hosting required due to public connectivity needs and my consumer level Internet connection at home, but that cost and flexibility is the tradeoff to have more self determination.
For this project, I have enlisted the help of Claude to test and build my design. As this is a fairly large effort with a lot of moving pieces, I have enlisted another Claude agent to blog about this effort, including publishing the current status of the project.
Note – this linked blog is AI generated. I call this out because I made a determination that I either have time to build the lab or to blog about it, but not both. As a lot of the engineering work is fed and processed through Claude, all of the notes and decisions are documented there and it is trivial to have the agent post to a separate blog.
One area of concern I have with this – I’m having an AI agent turn most of the wrenches on the software config of this environment, will I be able to support and effectively troubleshoot problems without an AI agent if it fails?
Part of the design is to make sure the most critical pieces are not load bearing. My files and photos will still reside in the file system on a NAS. Backups are stored both locally and with a cloud provider with my own encryption keys. If the key services fail, I still have my data and can access it (albeit without the fancy UIs and other enriched data)