The AI value chain, and the one layer no one else owns

Compute builds the substrate. Data feeds it. Models turn data into capability. Agents turn capability into action. KYE Protocol™ resolves whether that action may execute and become final. Models create capability. Agents create action. KYE™ resolves authority.

Your agents are already acting on systems you are accountable for. Most of the AI stack answers a build-or-connect-or-monitor question; KYE™ answers a different one, asked the instant an agent, model or workflow is about to do something consequential: can this entity act, on behalf of this principal, under this delegation, within this scope, against this policy, with this evidence, right now — and can the action become final? That question has no other home in the stack. KYE™ is the authority-finality layer.

See the ten governed stages

The nine layers — and KYE™'s zone

The chain runs from infrastructure up to feedback. KYE™ consumes state from the layers below it and emits evidence to the layers above it, but it occupies the boundary the others step over: the moment information becomes consequence.

Compute & infrastructure

GPUs, cloud, edge, inference runtimes. KYE™ consumes machine and runtime state; it does not compete here.

Data layer

Datasets, pipelines, vector stores, memory, lineage. KYE™ governs data and memory authority — may this agent use this data for this purpose? — it does not store the data.

Model layer

Foundation models, fine-tunes, evals, model cards. KYE™ consumes model capability and risk state; it does not train models.

Agent runtime & harness KYE™ zone

Planners, tool use, orchestration, MCP, task loops. KYE™ becomes the authority layer inside the harness — every tool call resolves against admissibility before it fires.

Application & workflow KYE™ zone

Copilots, vertical agents, ERP/CRM/HR workflows. KYE™ governs each workflow state transition and agentic action.

Execution & action KYE™ core

API writes, payments, refunds, code merges, document releases, robotic commands. KYE™'s core zone: should this action execute, and become final?

Evidence, audit & reporting KYE™ zone

Logs, replay, regulator and incident reports. KYE™ produces Evidence Packs™, Replay Proofs™ and regulator-ready reports — each a signed JSON payload derived from the decision, mapped to frameworks from SOC 2 to the EU AI Act, not reconstructed after it.

Governance, risk & assurance

Policy, compliance, oversight, certification. KYE™ turns governance into runtime authority and finality rather than a document review.

Feedback & learning

Drift detection, incident review, revocation, resilience loops. KYE™ feeds learning through finality failures, drift events, contestability and revocation (Reality Coupling™).

KYE™'s strongest claim is layers 4 through 7 — the span where an agent reaches for a tool, the action forms, executes, and leaves signed evidence. Upstream players build capability; downstream systems receive consequence. KYE™ is the authority and evidence layer in between.

Where KYE™ sits, exactly

A consequential act flows through KYE™ in a fixed order. Sensing becomes state capture; evaluation becomes Action Admissibility™; response becomes allow, deny, escalate or revoke; learning becomes evidence, finality failures and resilience-loop updates.

Model / Agent / Workflow→ Intent→ Action Admissibility™→ Authority Resolution™→ Delegation · Scope · Policy · Evidence→ Execution Control→ Tool · API · Payment · Document · Robot→ Evidence Pack™→ Authority Finality™→ Reality Coupling™ · Contestability · Revocation

The same spine, stage by stage, is the KYE Agentic Governance Lifecycle™ — this page is the zoom-out (the industry chain); that page is the zoom-in (every governed stage of a single act).

What KYE™ is — and what it is not

Name the layer precisely and you keep KYE™ out of four crowded categories it does not belong to.

KYE™ is the authority and evidence layer that determines whether an entity may perform a consequential action, and whether that action becomes final. Naming it precisely keeps it out of crowded categories it does not belong to.

KYE™ is not a model provider, an agent framework, a workflow engine, a data platform, an identity provider, a payment rail, a DLP tool, a SIEM, or a GRC dashboard.
KYE™ is the runtime authority and evidence layer for AI agents — the authority-finality layer of the AI value chain.

The category line, sharpened: project approval is not action authority. A deployment-readiness platform asks whether an AI project can ship; a security control plane makes agents visible as reportable assets. KYE™ governs the moment a verified entity actually tries to act — when AI stops being information and becomes consequence: before a payment, a refund, a code merge, a document release, a clinical report, a claim decision, an ERP write-back, a workflow transition, or a robotic command.