RUBI Edge Node
Enterprise intelligence at the edge
Persistent local intelligence — an optional deployment layer for organizations requiring greater privacy, local execution, offline resilience, lower latency and always-on operations. RUBI never requires it: without an Edge Node, work runs in the cloud, on desktop or on mobile.
Architecture
The Edge Node is not RUBI's brain
The intelligence lives in RUBI. The Edge Node is one runtime among several, chosen per task by the Execution Router.
RUBI
Persistent Operational Intelligence
│
RUBI ORCHESTRATOR
│
┌──────────────┼──────────────┐
CLOUD RUNTIME EDGE NODE USER DEVICES
└──────────────┼──────────────┘
RUBI INTELLIGENCEWhat exists today
The software layer is already running
Execution Router is live
Every RUBI task already passes through a deterministic router: task → policy → data classification → latency → network → compute → runtime. It decides Cloud, Edge, Desktop or Mobile and explains why. Try it below.
Nothing depends on hardware
No RUBI capability requires an Edge Node. When one is absent or offline, the router falls back to Cloud, Desktop or Mobile automatically — the work never stops.
Fleet management is ready
The Edge Fleet console in the admin area tracks each node's name, organization, location, status, CPU, memory, storage, agents, tasks, sync, last heartbeat and security state. It shows real registered devices — today there are none, and it says so honestly.
Offline queue and local-first state
RUBI's offline-first queue, persistent scheduler and heartbeats already run on user devices. The Edge Node extends this same architecture into a dedicated on-premises appliance.
Live demo · Execution Router
Where should RUBI run this?
This is the real router that ships with RUBI — the same logic, tested automatically. Pick a task and an environment; RUBI chooses the runtime and explains why.
RUBI decides
Run on cloud
- → Public data → Cloud
Designed for
What the Edge Node adds when present
- Always-on local runtime and scheduler
- Local agent execution and browser sessions
- Local connectors to enterprise systems
- Encrypted local memory and database cache
- Offline operation with cloud failover
- Local embeddings and speech processing
- Acoustic Intelligence and Smart Eyes near the source
- Local inference where the hardware supports it
- Remote management and signed automatic updates
Hospital
Authorized internal documents, selected patient-system connectors, local workflows, speech, monitoring, memory cache and approved private models. Only permitted workloads leave the building.
Factory
Local connections to machines, sensors, cameras, acoustic monitoring, inventory and maintenance — the Organizational Immune System keeps detecting signals with less dependence on cloud connectivity.
Office
Persistent company agents, local browser sessions, internal system connections, scheduled monitoring, a local file index and an offline queue — without employees' laptops staying on.
Security by design
Credentials never leave the vault
Credentials stay encrypted locally wherever the architecture allows, are never sent raw to AI models, and the credential vault is separated from the model context.
- Encrypted disk
- Secure boot where hardware permits
- Device identity
- Certificate-based authentication
- Remote revocation
- Signed automatic updates
- Audit logs
- Network segmentation
- Credential vault separated from AI context
- Isolated browser profiles per agent
