Distribution

Choose how you want to run AI Stays Local.

One runtime, four ways to run it. Distribution is invite-only during the private beta — which is an availability model, not an absence.

Desktop app

Private beta

Individuals and professionals who want the product, not the plumbing.

Guided installation
No terminal required
Local chat with your documents
Automatic model selection for your machine
Local memory
Controlled updates

Request desktop access

Docker

Private beta

Servers, NAS, workstations and technical users.

Isolated runtime image
Models mounted as a volume
Documents mounted read-only
CPU mode, or supported NVIDIA GPU
Linux ARM64 and x86-64 targets
Local API with health checks and resource ceilings

Request Docker access

CLI and API

Operational internally

Developers, power users, automation and integrations.

aisl command line
OpenAI-compatible local endpoint
Loopback binding by default
API authentication
Model registry and lifecycle management
Local documents with citations

Developer documentation

Enterprise

Private engagements

Organisations deploying on hardware they own.

Architecture qualification
Hardware and model validation
Deployment assistance
Security review support
Named support contact
Professional services

There is no public download button here because there is no public download. Builds go to approved private beta testers, and when a signed artefact is published it will appear with its checksum and its release notes.

How the Docker edition is put together Check your hardware first