What is NEES Core V3?

The distinction that matters

An AI agent can be technically capable of using a tool without having authorization for this specific actor, resource and action. NEES exposes a governance decision boundary outside the model's prompt. The owning application must still enforce that decision and check the real target.

A decision to ALLOW is not evidence that a tool ran or that the intended result was verified. Applications report actual outcomes and inspect supported evidence and receipts.

NEES is not an LLM or agent framework

• It does not replace your model, orchestrator, IAM system, or application-side authorization.
• It does not automatically intercept arbitrary tools, processes or requests outside a supported integration.
• It does not guarantee universal AI safety, compliance, cryptographic audit integrity or successful external execution.
• The underlying generic Core is private. Public integrations use documented Runtime and connector contracts.

Core engine versus hosted Platform

Core V3 is the reusable governance engine. The hosted Runtime exposes the submit → qualify → start → report contract. The Control Plane manages authorized governance configuration, the Dashboard supports developer onboarding and read-only assessment lookup, and supported connectors integrate participating applications.

Versions and historical names

Early Nainacore research used NEES in emotional-AI and memory contexts. Historical Core V2 and chat-oriented SDK material can still be found online. The current developer-facing action-governance product is NEES Core V3 / NEES Governance Platform. Historical package names must not be used as current V3 installation instructions.

Start with the public preview

A Free Developer account is self-service, without a payment method. Register a real supported action, configure and resolve published governance, and test from a trusted backend. This is a Developer Preview, not general availability.