Home integration
Home Assistant connects device and sensor information for lighting, climate, air quality, appliances, and security. Atlas brings selected information and controls into its household workspace.
The infrastructure behind Atlas
A home lab
built around
real life.
A working environment for connecting physical systems, household software, and AI-assisted workflows.
The lab has grown beyond its earlier dashboard and automation setup. Atlas now provides a shared household interface, while the services underneath it support device integration, energy visibility, data, and AI tools.
See Atlas and Galley Quest →The current foundation
Physical systems.
Software. Operational judgment.
Energy and data
Powerwall and solar information support energy visibility. InfluxDB and Grafana are part of the service stack for storing and exploring time-series data.
AI and supporting services
Ollama and Open WebUI support local AI tooling. The stack also includes search, notifications, travel services, and Galley Quest integration.
Today it all runs on Atlas, a single Windows 11 Pro machine. Home Assistant, InfluxDB, Grafana, Open WebUI, SearXNG, and ntfy run as Docker containers, Ollama serves local models, and scheduled tasks handle Atlas startup, health checks, and recovery.
Operate what you build
Visibility is part
of the design.
Atlas presents service observations, runtime status, and relevant device information in one Systems workspace. This helps separate the application interface from the infrastructure that supports it.
The lab also makes maintenance and recovery part of the work: understanding dependencies, observing changes, and verifying what a repair or restore actually recovers.
A reachable service and a useful result are different checks. That distinction guides how I evaluate the system.
Local where useful
Choose the right
boundary for each task.
Local compute is one part of the architecture. Device integrations, hosted data, and external AI or travel services can still require network access, accounts, or paid services.
I evaluate each workflow on its dependencies, privacy needs, and recovery options. Offline operation and reliability have to be demonstrated for the specific workflow.
Earlier work and current direction
Keep the lessons.
Improve the system.
JARVIS-5000, the wall dashboard that came before Atlas, is retired. Its code stays on GitHub as a reference. Atlas is the current household system, and Galley Quest shows how inventory, planning, and AI assistance can become a connected daily workflow.