Skip to main content
AI Inventory

Know which AI systems exist

Know which AI systems exist, what assets are connected, and what evidence is available.

AI Inventory is the starting point: select an AI system, open its AI Action & Access Map, review connected evidence, and see its assurance state.

See the evaluated Kestrel constellationExplore the Platform

From inventory to assurance

AI Systems
Know which AI systems and agents exist in your organization.
Connected Assets
See what assets, credentials, and tools are connected to each system.
System Constellation
Open the map to see what the AI can reach, change, and trigger.
Evidence
Review what evidence is established, partial, unknown, or not evaluated.
Assurance
See the bounded evaluation — ALLOW within evaluated scope, REVIEW, or BLOCK.

What AI Inventory provides

AI Systems
Catalog of AI systems and agents with type, provider, and context where available.
Connected Assets
Assets, credentials, and tools connected to each AI system where supported.
Evidence Availability
What evidence is established, partial, unknown, or not evaluated for each system.
Source / Provider Context
Repository and provider context where actually supported and connected.
System Constellation
Open the map for any system to see reach paths and potential consequences.
Assurance State
Assurance disposition and history where supported.

What AI Inventory does NOT do

It does not produce a global or aggregate risk score. Evidence and assurance are per-system and per-scope.

It does not imply universal provider completeness. What is not connected stays not connected.

It does not claim 24/7 monitoring of all systems. Monitoring is from connected runtime sources where supported.

It does not convert missing evidence into a pass. Unknown stays unknown.

Framework mapping

AI Inventory supports mapping to AI governance frameworks. Framework mapping supports assurance but does not equal assurance.

Colorado AI Act (SB 205)

Requires maintaining an inventory of AI systems with descriptions, intended uses, and risk classifications.

EU AI Act (Article 11)

Requires technical documentation and inventory of high-risk AI systems.

NYC LL144 (AEDT Registry)

Requires registration of automated employment decision tools.

NIST AI RMF

Recommends maintaining an inventory of AI systems as part of the Map function.

ISO/IEC 42001

Requires AI system inventory as part of the AI management system.

GDPR Article 30

Records of processing activities for automated systems processing personal data.

Framework mapping is not certification. HAIEC is not a certification body.

Usage Instrumentation

Drop-in wrappers around the OpenAI and Anthropic SDKs that automatically track usage and feed instrumentation data into HAIEC AI Inventory.

These are instrumentation packages, not security scanners. They do not scan code or detect vulnerabilities.

@haiec/openai

OpenAI SDK wrapper with automatic usage tracking.

$ npm install @haiec/openai

@haiec/anthropic

Anthropic Claude SDK wrapper with automatic usage tracking.

$ npm install @haiec/anthropic

Not security scanners. These packages instrument usage into HAIEC AI Inventory. For security checks, see Developer Security.

Start with the Map

See how AI Inventory connects to the AI Action & Access Map and assurance.

Explore the Sample MapSee the Platform