Architecture

Flexible private deployment from localhost to multi-floor enterprise networks.

Network reachability and compute capacity are separate: concurrent users still share available CPU/GPU, RAM, context and active workloads.

Deployment patterns

Choose the topology that matches the organization.

01 • Single Device / Localhost One user runs supported AI workloads on the same computer. Smallest network exposure; no LAN required.

02 • Single-Room Host / Private LAN One AI workstation serves approved devices on the same trusted room or office network.

03 • Multi-Floor Routed Host One stronger host serves approved floors or subnets through managed enterprise routing and firewall policy.

04 • Multi-Host Department Separate hosts handle departments, workloads or capacity requirements with independent model choices.

Private network security

Keep internal API services private.

Use localhost or approved private LAN/VLAN routes. Do not expose private RINC API services directly to the public internet, guest Wi-Fi or public port forwarding.

Credentials

Treat Local API and Robotics API keys as secrets and separate capabilities where policy requires.

Least privilege

Give users and applications only the capabilities they need.

Network controls

For multi-floor deployments, use managed routing, private subnets/VLANs and firewall restrictions.