Discovery
Map supported devices and their physical and logical relationships across the infrastructure estate.
iFabric
iFabric™ is Invences’ Autonomous Infrastructure Fabric, designed to bring discovery, observability, digital-twin evaluation and agentic operations into a shared infrastructure context.
Explore iFabric Enter the interactive tour See architectureiFabric / In focus
iFabric connects telemetry and topology across vendors to give agentic AI the operational context it needs. Its four modules link infrastructure discovery and observability with change simulation and governed remediation.
Inside iFabric / Interactive walkthrough
Explore the estate. Investigate an issue. Review a governed response.
Enter the walkthroughCapabilities
Map supported devices and their physical and logical relationships across the infrastructure estate.
Correlate telemetry, logs and KPIs across networks, GPUs, compute, storage, applications and Physical AI to detect anomalies and investigate causes.
Evaluate supported firmware, configuration and workload changes against a configured infrastructure model before production.
Use operational context to investigate incidents, propose supported responses and carry out permitted actions through configured workflows.
iFabric / How it works
Discovery, observability, digital twins and agentic operations. One shared context.
Find the estate. Reveal its dependencies.
Illustrative flows. Model coverage, integrations and permitted actions depend on the deployment.
Infrastructure inventory and operational signals feed discovery, observability and shared topology/event context. Operators can see the estate in Command Center without running a change or an action. A proposed change can be evaluated in the digital twin. nLLM supports agentic reasoning; configured actions require authorization, and fresh observations return to operators for outcome review.
Infrastructure inventory and operational signals feed discovery, observability and shared topology/event context. Operators can see the estate in Command Center without running a change or an action. A proposed change can be evaluated in the digital twin. nLLM supports agentic reasoning; configured actions require authorization, and fresh observations return to operators for outcome review.
This is a functional product architecture, not live telemetry or a physical deployment map. Discovery and observability enrich shared context; they are not exclusive alternatives or a compulsory serial pipeline. The twin evaluates supported proposed changes and is not a required hop for normal observation or every incident. The approved-response view illustrates the authorized path only. Missing evidence or denied authority stops execution for review. Authorization can use configured policy and approval integrations, including gFabric where deployed; nLLM advice does not grant permission. Tools act only on supported targets within scope. Fresh evidence establishes the observed result, including unresolved issues; the diagram does not assume successful recovery. Command Center presents topology, events, findings and outcome evidence. Timing is illustrative; repetition restarts the explanation.
Discovery. Inventory and relationships enrich shared topology and become visible to operators. Discovery does not require a twin evaluation or a remediation action.
Observability. Infrastructure signals become contextual health and anomaly evidence for operators. Shared discovery context informs observation; the view highlights the signal path.
Digital twin. A proposed change uses shared infrastructure context and supported model coverage. Evaluation results return to operators; the model does not itself authorize production deployment.
Approved response. This view illustrates an approved agentic response: operational evidence grounds nLLM reasoning, authorization limits the action, configured tools execute it, and fresh signals return to Command Center for outcome review.