Waiting for your Integration Hub
Your network. One clear view.
Connect field systems. Turn measured signals into informed action.
From antenna to core.
Native routing policies and explicit physical commissioning assumptions for the selected site.
Multi-vendor policy workbench
Compile the same bounded routing intent into platform-specific syntax.
Radio & antenna commissioning
Generic planning values. Use the actual equipment datasheet for a field design.
Site equipment register
Control boundaries remain visible for every part of the site.
Failure is rarely a single fault.
Inject a case, collect evidence, review a scoped plan and prove recovery. Cases compound with the current run.
Resolution register
Three stable service observations, current visibility, diagnostics and a verified change are required.
Scheduled fault playbooks Controlled recovery timelines
Fault scenario library
Schedule failures and recoveries against the current lab state.
Schedule an experiment
Faults execute at a precise simulated time.
For down/up switches use 1 = down, 0 = restored. Other faults use the displayed unit.
Experiment timeline
Pending, applied and cancelled faults remain in the audit trail.
Network traffic
Offered vs delivered throughput · all sites · Mbps
AI copilot
Evidence → analysis → reviewed action
Connect the Hub to run provider analysis. The monitoring brief uses calculated model evidence.
Site health history
Transport + radio + power · worst condition per time bucket
Regional capacity
Offered load / provisioned budget
Telemetry pipeline
Dependency state & consumer offsets
Every site. Every dependency.
Regional health derived from calculated transport, radio and power conditions.
Open fleet telemetry ledger
Engineering controls Topology · fault injection · selected-site trends
Transport topology & selected-site trends
Dual paths, six aggregation domains and a shared core. Select a site below to investigate.
Selected site / fault controls
Faults change the environment. Reviewed jobs change configuration.
Advance the clock to calculate telemetry, detect incidents and verify changes. A down link takes three ticks to be detected by routing.
Investigate the evidence
Rule-based hypotheses from model observations. Validate with a diagnostic probe.
Inspect before changing.
Probes are bound to site conditions and expire for approval after 30 ticks.
Access & mobile sessions DHCP · slicing · handover
Access services & mobile admission
DHCP pool lifecycle, subscriber policy, registration and modeled PDU sessions.
Public fixtures demo-1…demo-6. Weighted slice allocation shares 40 Mbps per cell. Handover interrupts the modeled user plane for two ticks and preserves the session address when successful.
Delivery, offsets & evidence storage
Stop a dependency, advance time, then restore it to inspect backlog recovery.
Review before execution
Operator proposes. Reviewer approves. Operator executes. Three observations decide verification or rollback.
Reproducible operational evidence
Inspect a command, a state transition or a stored modeled artifact. Export preserves the complete run.
Model rules & assumptions
WORKSPACE SETUPConnect your workspace to get started
Unique identities across connected sources
Evidence backed, consecutive confirmations
Durable delivery · Kafka & workflow
Field integrations
Monitoring, inventory, and network labs.
From signal
to next step.
Groq interprets evidence. OpenAI Decisions selects a next step for operator review.
Operational pipeline
Field → event → detection → decision
Recent incidents
Prioritize with severity and measured evidence.
One identity across your network
NetBox supplies inventory and maintenance context. Aliases in
config/connectors.yml map Zabbix, LibreNMS, GNS3, and
EVE-NG devices to a shared identity.
Provider & policy
Groq → OpenAI Decisions → review.
An incident must be open, evidence current, and the device outside maintenance. Decision confidence must meet your threshold. Provider failures close the review gate.
Analyze open incidents
Select an incident to request an AI analysis.
Decision audit trail
Models, probabilities, evidence, and policy checks in one record.
Platform health
Readiness probes and worker heartbeats · cached for 30s
Network regions
Inventory footprint and open incidents
Delivery ledger
Latest 100 events · acknowledgement, retries, and backoff
Device inventory
Source observations. Shared device IDs count once in the overview.
Network lab topology
Nodes and connections reported by GNS3 / EVE-NG.
Enable network intelligence
Set the values in .env, then restart the
Integration Hub. Provider credentials stay on the server.
AI_ENABLED=true GROQ_API_KEY=... OPENAI_API_KEY=... GROQ_MODEL=openai/gpt-oss-20b DECISION_MODEL=gpt-6-luna
Analysis runs when an operator selects Analyze with AI. Selected incident facts are sent to both providers.
Connect your operational workflow
Connect n8n, an AWX gateway, or your operations backend through an endpoint that accepts Idempotency-Key.
WORKFLOW_ENABLED=true WORKFLOW_WEBHOOK_URL=... WORKFLOW_WEBHOOK_TOKEN=...
Strata queues a review request. Your operational workflow handles approval and runbook execution.