Aevum Cadence AI

Local hosting

Local hosting means the assistant runs on your Linux machine. Data stays on your disk. You hatch it yourself and name it. There is no seeded house persona, and there is no Aevum-hosted cloud home in 0.1.

Download the signed AppImage or deb from the home page. Apt install is documented there too.

What you get in 0.1

Cloud hosting is not offered. There is no Cadence Cloud signup, no WorkOS login for this product, and no metered hosted-runtime SKU.

Native process

The packed app starts the assistant as a local process. No container required. Fastest setup, lowest latency, full access to this machine under permissions (sandbox bash vs confirm-gated host_bash).

The tradeoff is isolation: it is your user session. Confirm host writes. Do not treat a hatch as unattended admin on a shared server.

Docker

The open-source tree can run the assistant in Docker (separate volumes for gateway security and CES). That is a power-user path, not the 0.1 download. The AppImage/deb sit is native.

Comparison

Packed Linux (0.1) Cadence Cloud
Where it runs Your computer Not offered
Data On disk in the hatch directory Not offered
Model Hive or BYO OpenAI-compat Not offered
Availability While this machine is awake and the app is running Not offered

When local is the right choice

The availability tradeoff

Schedules, watchers, and heartbeats run only while the assistant process is up. Sleep or shut down, and they pause. Missed one-shots fire on next start; recurring jobs resume from now. See Scheduling.

If you need 24/7 without keeping a box awake, that is not a Cadence Cloud SKU today. Run the packed app on a machine you leave on, or wait for a later hosting option. Do not assume Aevum will host it for you in 0.1.

Updates

Packaged Linux uses the linux-electron feed. In the app: Help → Check for Updates. Or install the newer AppImage / deb from the home page.

Privacy of a local hatch: Privacy and data.