Shared awareness
Work from the same operational picture.
Connected stations contribute live data without losing their local context.
OpCopilot Server
One intelligence layer over the whole organisation.
A data-centre-class OpCopilot instance that aggregates and analyses data from every station at once — one unified picture across all sectors, for 50+ concurrent users. On-premise or private cloud.
Shipping · customer-deployed · human-in-the-loop
Server mesh / 04
Correlation active
Aggregation is the point
See the pattern no single station can see.
Server correlates events, history, and decisions across every connected station. The result is one operational picture for the whole organisation, not a larger version of a station console.
Every connected station
One shared picture
Cross-sector correlation
Operator-approved action
Scale and collaboration
Multiple watches. Multiple agencies. One memory.
Rooms, reports, findings, and decisions accumulate around the same events without collapsing agency or operator accountability.
Concurrent operations
Serve 50+ users at once.
Multiple watches and agencies can investigate, report, and coordinate concurrently.
Decision memory
Carry context across shifts.
Sources, questions, findings, and decisions remain available to the next watch.
Runs on
Data-centre compute selected by the customer.
The same containerised software lands on the substrate the organisation's policy requires.
- Accelerators
- Supported accelerators include data-centre-class NVIDIA H20, H100, and H200 GPUs.
- Customer-controlled substrate
- Delivered as infrastructure-as-code onto bare-metal Linux, a customer-controlled cloud tenancy, or a private-cloud equivalent; the customer selects and owns the substrate.
- Private infrastructure
- Containerised microservices deploy on bare metal or private cloud infrastructure.
- Deployment boundary
- Runs on-premise or in a customer-selected private cloud.
Provisioned as code
Software, not a snowflake installation.
OpCopilot Server is provisioned as infrastructure-as-code onto bare-metal Linux, EC2, or a private-cloud equivalent.
The same build is reproducible on whichever substrate policy requires.
Reproducible rebuilds
No snowflake handover
Portable sovereignty tiers
No re-implementation
What runs on it
The complete OpCopilot capability layer.
Server runs the same skills, MCP connectors, reasoning model, and human approval flow as Edge and Plugin.
Sense
Aggregate every station.
Feeds and systems contribute to a unified, cross-source operational picture.
Reason
Apply one governed ConOps.
Versioned skills bring doctrine, thresholds, and escalation paths into the reasoning.
Act with the operator
Keep decisions accountable.
Every recommendation is traceable and every action remains subject to human approval.
Operations
A deployment the customer can operate.
Delivery covers installation, security hardening, operator training, lifecycle practices, and software updates within the selected environment.
- Installation on customer-selected infrastructure
- Security hardening for the deployment boundary
- Operator training and operational adoption
- MLOps lifecycle management and updates
OpCopilot editions