DigitalOceanLive validated
Action completion + resource state
A dependent action remained held until both the asynchronous operation completed and a separate resource read proved the machine was active.
Technical evidence
Evidence, not a concept slide.
The core Varqo pattern has been exercised against real providers and a real tool-using agent: prepare proof, bind runtime identity, verify independently, hold continuation, and release only after the required state is proven.
Live provider proofs
Different protocols. The same integrity invariant.
Varqo has verified same-resource, cross-resource, runtime-URL, and POST/batch provider topologies while preserving provider-native lifecycle state.
Microsoft AzureLive validated
Runtime-discovered verifier
Varqo safely bound and executed a verifier URL that became available only after the real write response.
AWS CodeBuildLive validated
POST / batch verification
A runtime build identity was bound into an independent batch read and polled from in-progress to its provider-native terminal state.
Tool-using AI agentLive validated
Controlled dependent execution
A real agent attempted to complete its workflow before proof was ready. Varqo held that continuation and released it only after verification.
Evidence scope: These are controlled live proofs of the integrity kernel and local control boundary. They establish the technical pattern without claiming broad production deployment or enterprise-scale coverage.
What the experiments establish
The kernel is provider-neutral where it matters.
The implementations differ, but the control decision remains tied to independently observed state and an explicit outcome requirement.
Automatic verifier planning
High-level task intent, provider metadata, the actual operation, and runtime context compile into executable proof.
Runtime identity binding
Plans can be prepared before a write and safely bound to the real action or resource identity returned afterward.
Composite outcome proof
Action completion alone cannot release a workflow when the downstream action also depends on resulting resource state.
Persistent low-latency plans
Semantic preparation can move out of the live path while parameter binding remains measured in fractions of a millisecond.
Early design partnership
Talk to Varqo Bring us a consequential agent workflow.
We are looking for teams operating agents with cloud or enterprise permissions to help validate the next integration and resilience milestones against real constraints.