Evidence infrastructure for autonomous computing

AI will not be governed by logs alone.

RootCompute connects what infrastructure independently observed with what information claims about its origin — creating a clearer evidence chain across execution, data, provenance and, ultimately, autonomous action.

The shift

AI is already generating information from information that was itself generated by AI.

“AI-generated” is no longer specific enough. Organizations increasingly need to establish which system produced an output, what execution context surrounded it, what information was associated with it, and what happened afterward.

RootCompute platform

A signed evidence graph, not another dashboard.

RootCompute is being built to separate evidence by source and trust boundary, then correlate and bind it into records that other systems can inspect, verify and query.

01

Observed

Infrastructure-derived evidence from a supported observation boundary.

02

Declared

What the application, model, operator or provenance record says happened.

03

Bound

A protected relationship between observation, provenance, time and workload identity.

04

Unknown

Explicit gaps and unsupported conclusions instead of false certainty.

Product planes

See the workload. Trace the information. Audit the action.

RC RuntimeExecution and integrity evidence from supported compute environments.
RC DataSelected data-in-use sensing and attribution inside reconstructed CPU address spaces.
RC ProvenanceGeneration records, watermark signals and compatible provenance references.
RC AgentFuture evidence for consequential autonomous actions and tool use.
OBSERVEBINDVERIFY

Evidence producers

Rooted in execution. Proven across information.

inVtero and XAMMY are distinct technical foundations. RootCompute is the proposed evidence layer that normalizes, correlates, signs, stores, queries and exposes what those systems can establish.

Evidence, not assertions

RootCompute adds observation to provenance.

The goal is not to replace host logs, runtime agents, C2PA, infrastructure telemetry or existing security products. It is to add independently observed evidence where possible, bind it to provenance, and keep inference and unknowns explicit.

Machine

What actually ran?

Reconstruct supported runtime state without assuming the guest software is the only trustworthy witness.

Information

Where did this come from?

Carry generation evidence, detector context, content identity and compatible provenance references forward.

Action

What did the agent do?

Extend the same evidence-first architecture toward autonomous tool use and consequential machine actions.

Security Objectives / RootCompute

Autonomous computing needs defensible machine evidence.

RootCompute is a pre-release combined offering. Observation coverage depends on the acquisition path, supported representations and configured access. Provenance is evidence of origin or association, not a substitute for independent quality, truth or policy evaluation.