Should your family AI assistant run at home or in the cloud?
Choose a home Mac if Lumstead needs to control devices on your home network or run a model there. Choose hosted access if you want the service to run the server for you. Hosting sets where your records live. Your model choice sets where your AI requests go.
By the Lumstead team · Product previewWhich tasks need a home server?
A shared grocery note or meal plan can live in either setup. Controlling a supported television on your home network requires a local Lumstead server. The hosted service cannot reach the TV. If you want both a film list and TV control, choose a home server. If you only want the list, either setup can hold it.
What happens when I leave home?
A home server needs power, a network connection, and a secure way for your phone to reach it. Check access from mobile data before you rely on it for a shopping trip. If the Mac sleeps, your phone loses access to the service. You also manage updates and backups. Keep the backup on separate encrypted storage and check that you can restore it.
With a hosted server, your phone connects to the service over the internet and the operator maintains the machine. Lumstead is taking waitlist signups for hosted access.
Can I keep model requests on my Mac?
Yes. Run Ollama on the home Mac, use the address set by the server owner, and select a downloaded model that supports tool calls. Ask it to read a sample note and compare its answer with the note. The separate “Your inference server” connection requires a public HTTPS endpoint. Use the Ollama connection for a model on your home network. Even with a local model, production sign-in uses email and needs internet access.
What costs should we compare?
For a home Mac, count the machine, electricity, backup storage, and the time someone spends maintaining it. Add provider charges if you choose remote AI. When comparing a hosted offer, check what the service includes and who pays for model use. With a personal API key, the provider bills the key owner.