title: “Schema Architecture” description: “How AIGO machine-readable schemas represent AI governance entities, records, relationships, and lifecycle information.”

Schema Architecture

The AIGO Framework includes a machine-readable schema layer that provides structured representations of AI governance entities, records, relationships, and lifecycle activities. The schema layer translates the conceptual AIGO Framework into structures that can be stored, exchanged, validated, integrated with governance platforms, and used by supporting automation. The schemas should therefore be understood as an implementation representation of AIGO governance information, rather than as a replacement for the human-readable framework.

Purpose

The AIGO schema layer supports: * structured governance records * consistent data representation * traceability * interoperability * validation * governance-system integration * evidence management * machine-readable relationships * evaluation and automation * lifecycle management An organization can use the schemas as the structural foundation for implementing AIGO in a governance application, database, registry, workflow platform, or other technical environment.

Conceptual model

AIGO separates three related layers: Framework → Governance Records → Machine-Readable Representation The Framework defines the governance concepts, principles, domains, lifecycle, risks, controls, and decision model. Governance records represent the information produced while applying those concepts. Schemas define how those records can be represented structurally. For example: AIGO Framework defines that AI systems should be governed. AI System Governance Record contains information about a particular AI system. AI System Schema defines the machine-readable structure used to represent that record.

Schema families

The canonical AIGO v0.1 schema family includes: 1. AI System 2. Risk 3. Control 4. Assessment 5. Approval 6. Monitoring 7. Incident 8. Change 9. Assurance 10. Evidence 11. Management Review 12. Improvement 13. Retirement 14. Governance Each schema represents a defined governance object or record type.

Schema architecture

The canonical schema architecture is organized as:

AIGO Governance
│
├── AI System
│
├── Risk
│
├── Control
│
├── Assessment
│
├── Approval
│
├── Monitoring
│
├── Incident
│
├── Change
│
├── Assurance
│
├── Evidence
│
├── Management Review
│
├── Improvement
│
├── Retirement
│
└── Governance

These are not isolated data structures. They form a connected governance information model.

Governance relationship model

A simplified lifecycle relationship is: AI system → classification → risk → control → evidence → assessment → approval → monitoring → assurance → improvement → retirement Additional relationships can occur throughout the lifecycle. For example: * an incident may create a new risk * a change may trigger reclassification * a risk may require additional controls * a control assessment may require additional evidence * an assurance finding may create an improvement action * an improvement may require a change to an existing control * retirement may require preservation of governance evidence The schema layer allows these relationships to be represented without requiring all governance information to exist in a single record.

Core entity: AI system

The AI system is a central governed entity within AIGO. Other governance records can reference or relate to the AI system. For example:

AI System
   │
   ├── Classification
   ├── Risks
   │    └── Controls
   │          └── Assessments
   │                └── Evidence
   │
   ├── Approvals
   ├── Monitoring
   ├── Incidents
   ├── Changes
   ├── Assurance
   ├── Improvements
   └── Retirement

This structure allows an organization to maintain specialized governance records while preserving traceability back to the governed AI system.

Governance records

A governance record is a structured representation of information generated during governance activities. Examples include: * an AI system registration * a risk record * a control record * a control assessment * an approval decision * a monitoring record * an incident record * a change record * an assurance record * an evidence record * an improvement record * a retirement record Schemas define the expected structure of these records.

Relationships and traceability

AIGO relies on relationships between governance records to maintain traceability. For example:

AI System
    ↓
Risk
    ↓
Control
    ↓
Evidence
    ↓
Assessment
    ↓
Approval
    ↓
Monitoring
    ↓
Assurance
    ↓
Improvement

The actual relationship structure may be more complex and can include multiple risks, controls, assessments, evidence records, approvals, incidents, changes, and assurance activities. The important principle is that governance information should remain traceable to the relevant governed entity and decision.

Schema identifiers and relationships

When implementing the schemas in a technical system, organizations should preserve stable identifiers for governed records. For example, an implementation may use identifiers such as:

AI-000123
RISK-000456
CTRL-000789
ASSESS-000321
APP-000654

The specific identifier format is an implementation decision unless otherwise defined by the applicable AIGO schema. What matters is that records can be uniquely identified and related.

Schema validation

Schemas can be used to validate the structural integrity of governance records. Validation can determine whether a record conforms to the expected machine-readable structure. Examples include checking: * required fields * field types * allowed values * structural relationships * object structure * array structure * identifier formats * schema references Schema validation is a structural quality mechanism. A record passing schema validation does not automatically mean that the underlying governance decision is correct, effective, legally compliant, or sufficient.

Schema documentation

Each canonical schema is accompanied by schema documentation explaining its purpose and structure. The canonical schema documentation provides additional information about: * schema purpose * objects * fields * relationships * required information * enumerations * implementation considerations * governance meaning Users implementing AIGO should read the relevant schema documentation together with the corresponding framework and implementation guidance.

Using schemas in an implementation

Organizations can implement the AIGO schemas in different technical environments. Possible implementations include: * relational databases * document databases * governance platforms * GRC platforms * API services * workflow systems * AI registries * internal governance applications * JSON-based repositories * data warehouses * evidence-management systems AIGO does not require a single technology architecture. The schemas provide a common structural representation that can be adapted to the organization’s technical environment.

Recommended implementation sequence

Organizations implementing the machine-readable layer should generally proceed in stages.

1. Establish the AI system record

Create the structured representation of each governed AI system.

2. Establish governance relationships

Connect systems to relevant risks, controls, assessments, approvals, monitoring, and other records.

3. Establish identifiers

Ensure governed records have stable identifiers that allow traceability.

4. Implement validation

Validate records against the applicable schemas.

5. Implement lifecycle workflows

Connect schema records to the organization’s governance processes.

6. Implement evidence relationships

Ensure governance decisions and assessments can be connected to supporting evidence.

7. Implement evaluation where appropriate

Where machine-readable rules are available, governance records can become inputs to automated or semi-automated evaluation.

8. Implement reporting and assurance

Use structured records to support governance reporting, monitoring, assurance, and management review.

Schemas and governance workflows

Schemas should not be implemented independently of governance procedures. A practical implementation connects: procedure → workflow → governance record → schema → validation → evidence → decision For example:

AI Registration Procedure
        ↓
Registration Workflow
        ↓
AI System Record
        ↓
AI System Schema
        ↓
Schema Validation
        ↓
Governance Review

This allows the technical implementation to remain aligned with the actual governance process.

Schemas and machine-readable evaluation

The schema layer also provides structured inputs for the AIGO evaluation architecture. A simplified relationship is: Governance Records → Rules → Evaluation → Findings For example:

AI System Record
       +
Risk Record
       +
Control Record
       +
Evidence
       ↓
Governance Rule
       ↓
Evaluation
       ↓
Result / Finding

The evaluation layer should remain logically separate from the underlying governance records. This allows rules to evolve without changing the identity or historical record of the governed entity.

Schemas and evidence

Evidence is an important part of the machine-readable governance model. Governance decisions should be capable of being connected to evidence where appropriate. Examples include: * test results * approval records * monitoring outputs * assessment results * policies * procedures * technical documentation * audit records * incident records * assurance findings The evidence relationship helps establish why a governance conclusion was reached and supports subsequent review or assurance.

Schemas and lifecycle management

The schema family supports governance throughout the AI system lifecycle. A simplified lifecycle representation is:

Register
   ↓
Classify
   ↓
Assess
   ↓
Control
   ↓
Approve
   ↓
Operate
   ↓
Monitor
   ↓
Change / Incident
   ↓
Assure
   ↓
Improve
   ↓
Retire

Governance records can be created or updated as the AI system progresses through these stages. This enables an organization to maintain a historical governance record rather than treating governance as a collection of disconnected documents.

Canonical schemas

The canonical JSON schemas are maintained under: schemas/ The canonical schema architecture is: schemas/00-AIGO-Schema-Architecture-v0.1.md The schema documentation is organized into the following areas: * schemas/01-ai-system/ * schemas/02-risk/ * schemas/03-control/ * schemas/04-assessment/ * schemas/05-approval/ * schemas/06-monitoring/ * schemas/07-incident/ * schemas/08-change/ * schemas/09-assurance/ * schemas/10-evidence/ * schemas/11-management-review/ * schemas/12-improvement/ * schemas/13-retirement/ * schemas/14-governance/

Schema documentation pages

The Mintlify documentation provides dedicated pages for each canonical schema family: * AI System Schema * Risk Schema * Control Schema * Assessment Schema * Approval Schema * Monitoring Schema * Incident Schema * Change Schema * Assurance Schema * Evidence Schema * Management Review Schema * Improvement Schema * Retirement Schema * Governance Schema These pages should be read together with the corresponding framework and implementation guidance.

Implementation principle

The schemas should be treated as a structured governance representation, not as the governance framework itself. The framework defines what should be governed. The implementation guidance explains how organizations can perform the governance activities. The schemas define how governance information can be represented. The rules and evaluation layer defines how selected governance conditions can be evaluated. The validation layer helps verify the integrity of the resulting artifacts. Together these layers provide the foundation for implementing AIGO in both human-operated and technology-supported governance environments.

Version

The canonical schema family described here is based on the AIGO Framework v0.1 schema baseline. Subsequent implementation-layer capabilities should be interpreted according to their individual version and repository status and should not silently be treated as changes to the normative framework baseline.

​title: “Schema Architecture” description: “How AIGO machine-readable schemas represent AI governance entities, records, relationships, and lifecycle information.”

​Schema Architecture

​Purpose

​Conceptual model

​Schema families

​Schema architecture

​Governance relationship model

​Core entity: AI system

​Governance records

​Relationships and traceability

​Schema identifiers and relationships

​Schema validation

​Schema documentation

​Using schemas in an implementation

​Recommended implementation sequence

​1. Establish the AI system record

​2. Establish governance relationships

​3. Establish identifiers

​4. Implement validation

​5. Implement lifecycle workflows

​6. Implement evidence relationships

​7. Implement evaluation where appropriate

​8. Implement reporting and assurance

​Schemas and governance workflows

​Schemas and machine-readable evaluation

​Schemas and evidence

​Schemas and lifecycle management

​Canonical schemas

​Schema documentation pages

​Implementation principle

​Version