AI governance + assessment · NSAG applied
GAPI — AI Agent Governance Infrastructure
Framework and readiness-prototype concept · release-blocked
Evaluate agent architecture through nervous-system-aware governance dimensions and make governance evidence legible to enterprise users.
Controlled visual demonstration
Current live interface
This sandboxed preview shows the current public interface. Open the full site to use the build in its intended window.
What I noticed
AI-agent assurance is frequently reduced to model capability, output accuracy, or general ethics language even though deployed conversational systems also shape attention, cognitive load, trust, choice architecture, and emotional state. Organizations need a way to document those influence mechanisms, connect them to human oversight, and monitor how they change after deployment.
What I researched
GAPI extends NSAG modules M2 and M11 into six assessment dimensions: influence-profile documentation, cognitive-load governance, transparency and explainability, human oversight and override, demographic and population impact, and post-deployment monitoring. The concept places those dimensions alongside public signals such as the NIST AI Risk Management Framework and FTC enforcement while proposing an infrastructure-level implementation rather than a periodic policy-only review.
What exists
A six-dimension NSAG governance assessment and certification-mark concept for AI agents.
Who it serves
AI product teams, enterprise buyers, governance leaders, and assurance functions.
What it demonstrates
AI-agent governance, assessment design, certification architecture, and translation of institutional theory into product controls.
Current evidence
The live NSAG-applied concept verifies the six-dimension assessment architecture, readiness diagnostic, signal-versus-noise guidance, glossary, hypothetical transformations, tiered engagement model, and proposed Agent Governance Certification mark and MCP-server direction. It does not verify a functioning governance API or MCP server, completed technical integration, customer assessment, audited model, certification decision, monitoring system, buyer acceptance, or safety outcome. The mark is a proposed NSAG program—not independent accreditation, regulatory approval, an industry standard, or a guarantee that an agent is safe, lawful, unbiased, or suitable for a particular use. Legal, scientific, security, demographic-impact, and assurance claims require domain review and empirical validation.
Verified build record
What is actually running, and what it is not.
LiveLive
The problem. Organisations deploying AI agents are asked for an inventory, a risk assessment and an evidence trail at the moment a regulator or a customer asks, which is the worst possible moment to start assembling one.
What it does. Works through AI system inventory, risk triage, controls, evaluation, incidents and human oversight, and exports the result as an evidence pack. Everything stays in the browser.
Built on. A static browser application with local persistence, plus privacy and terms pages, served from Vercel.
What it does not establish. Local-first issue-spotting. Not a compliance certification and not a legal assessment.
This record is generated from the same inventory as the complete build index: the production URL was checked at the last regeneration, and the status line above repeats what the build reports about itself rather than restating an intention.