Private AI on the hardware you already own.
A self-contained AI node: local inference over GGUF models, running in-process on this machine. With a Professional licence it can also hand a request that is too large for this box to another machine you own — one you add by address and allow by name, on your own network.
With no machine added, there is no path off this device: the plugin opens no listener and binds no port, and the only inference path is a local socket.
There is no discovery. You type a private address, you allow that machine by name, and only then can a request go to it — one function decides, so "did we ask" is checkable in one place.
Small requests stay here. A request leaves only when it is too large to run well on this box, or when you name a destination for it yourself.
Every peer added, allowed, revoked and used is written to a tamper-evident audit ledger — the destination and the size, never the prompt or the reply.
What each tier unlocks, generated from the same catalogue the application gates against — so the feature list here is exactly what the software enforces. Pricing is announced at launch.
| Plan | Price | Includes |
|---|---|---|
| Community | Included |
|
| Professional | $99/yr |
|
| Team | $299/yr |
|