​

Monitoring Schema

The AIGO Monitoring Schema provides the machine-readable structure for recording ongoing monitoring activities, observations, indicators, results, and governance responses. Monitoring provides the operational feedback mechanism through which an organization can detect changes and determine whether additional governance action is required.

​

Purpose

The Monitoring Schema supports consistent recording of:

• what is being monitored
• why it is being monitored
• the monitoring scope
• observed conditions
• indicators and measurements
• thresholds or criteria
• monitoring results
• identified signals or findings
• responsible parties
• required actions
• escalation
• follow-up activities
• supporting evidence

This allows monitoring information to become part of the governed AI system lifecycle rather than remaining only in operational dashboards or informal records.

​

Role in the AIGO architecture

Monitoring provides ongoing feedback after governance decisions have been made and throughout AI system operation. A representative lifecycle relationship is: AI system → risk → controls → assessment → approval → operation → monitoring → reassessment / action Monitoring can therefore connect operational activity back into governance. Monitoring results may lead to:

• risk reassessment
• control reassessment
• change management
• incident management
• approval review
• assurance
• corrective action
• continual improvement

​

What monitoring represents

Monitoring can address both the AI system and its governance environment. Depending on the system and context, monitoring may consider:

• system performance
• reliability
• safety indicators
• security indicators
• data quality
• model behavior
• output quality
• risk indicators
• control performance
• incidents
• changes
• policy adherence
• approval conditions
• operating conditions
• stakeholder impacts
• emerging risks
• assurance findings

The appropriate monitoring scope depends on the characteristics, risks, intended use, and governance requirements of the AI system.

​

Monitoring scope

A monitoring record should establish what is being observed and within which context. The scope may identify:

• AI system
• model or component
• deployment
• use case
• control
• risk
• governance condition
• approval condition
• operational environment
• defined monitoring period

Clear scope is important because monitoring results should not be interpreted outside the conditions under which they were collected.

​

Indicators and measurements

Monitoring may use defined indicators to detect relevant changes or conditions. Indicators can include:

• performance metrics
• risk indicators
• control effectiveness indicators
• incident frequency
• error rates
• threshold breaches
• drift indicators
• operational measures
• compliance indicators
• governance process measures

Where quantitative thresholds are used, the monitoring implementation should preserve the relationship between the observed value, the defined threshold or criterion, and the resulting governance response.

​

Monitoring results

Monitoring results may indicate:

• normal operation
• threshold reached
• threshold exceeded
• emerging concern
• control weakness
• increased risk
• incident
• material change
• governance exception
• reassessment required

The exact result vocabulary depends on the applicable AIGO implementation and machine-readable definitions. A monitoring result should be interpreted together with its scope, indicators, evidence, and context.

​

Relationship to risk

Monitoring provides an important input to ongoing risk management. A typical relationship is: risk → risk indicator → monitoring → observation → reassessment → risk treatment Monitoring can identify changes in:

• likelihood
• potential impact
• exposure
• affected stakeholders
• operating conditions
• control effectiveness
• emerging threats
• known failure modes

Monitoring should therefore provide a mechanism for detecting when an existing risk assessment may no longer represent the current operating environment.

​

Relationship to controls

Controls can be monitored to determine whether they continue to operate as intended. For example: control → monitoring indicator → observation → finding → control assessment Monitoring does not necessarily replace formal control assessment. Instead, monitoring can identify conditions that trigger a more detailed assessment. Repeated monitoring signals may also indicate that a control requires redesign, additional evidence, remediation, or assurance.

​

Relationship to approval

Approval conditions can become monitoring requirements. For example, a conditional approval may require:

• additional performance monitoring
• restricted-use monitoring
• periodic evidence
• defined threshold monitoring
• management review
• periodic reassessment

This creates a relationship such as: approval condition → monitoring requirement → monitoring result → governance action If an approval condition is not satisfied, the result may require escalation or review of the underlying approval.

​

Monitoring and incidents

Monitoring can provide an early detection mechanism for incidents. A monitoring result may trigger incident management when a defined condition indicates an actual or suspected incident. The relationship can be represented as: monitoring signal → incident assessment → response → evidence → corrective action Not every monitoring anomaly is necessarily an incident. Organizations should define the criteria used to determine when monitoring observations require incident handling.

​

Monitoring and change management

Monitoring can identify changes that were not previously recorded through the formal change-management process. Examples include:

• unexpected changes in system behavior
• changes in operating conditions
• changes in data characteristics
• changes in stakeholder impact
• changes in risk indicators
• changes in control performance

These observations may trigger investigation and, where appropriate, formal change management.

​

Monitoring and assurance

Monitoring provides operational evidence that can support assurance activities. Assurance may examine whether:

• monitoring requirements were defined appropriately
• monitoring was performed as required
• indicators were meaningful
• thresholds were appropriate
• monitoring results were acted upon
• significant findings were escalated
• evidence was retained
• monitoring supported governance decisions

This distinguishes operational monitoring from independent or management assurance over the monitoring process.

​

Monitoring frequency

Monitoring frequency should be proportionate to the AI system and governance context. Factors that may influence frequency include:

• system risk
• deployment scale
• potential impact
• system volatility
• rate of change
• incident history
• control significance
• regulatory requirements
• approval conditions
• organizational policy

Some indicators may require continuous or near-real-time observation, while others may be reviewed periodically. AIGO does not require a single monitoring frequency for all AI systems.

​

Escalation and governance response

Monitoring should have defined response pathways for significant results. Potential responses include:

• investigation
• risk reassessment
• control assessment
• corrective action
• incident management
• change management
• approval review
• suspension
• additional monitoring
• assurance
• management escalation
• continual improvement

The response should be proportionate to the significance of the observed condition.

​

Evidence

Monitoring records should maintain appropriate evidence supporting significant observations and conclusions. Evidence may include:

• system logs
• metrics
• monitoring reports
• test results
• alerts
• dashboards
• incident records
• control records
• approval records
• assessment results
• other operational or governance artifacts

Evidence should be retained according to applicable organizational requirements.

​

Relationship to machine-readable evaluation

Monitoring records can provide inputs to AIGO’s machine-readable governance evaluation layer. The architecture separates: governance records → evaluation rules → evaluation → result Monitoring data can therefore be evaluated against defined governance conditions without embedding evaluation logic directly into the Monitoring Schema. For example, evaluation may determine whether:

• required monitoring exists
• monitoring is current
• a threshold has been exceeded
• required monitoring evidence exists
• an approval condition is being monitored
• a required governance response has been initiated

Automated evaluation can support detection and workflow but does not automatically determine the appropriate organizational response in every circumstance.

​

Machine-readable implementation

An organization implementing AIGO may use the Monitoring Schema as the structural contract for monitoring records. A monitoring implementation could:

1. identify the monitored AI system or governance object
2. define the monitoring scope
3. define indicators or conditions
4. collect observations
5. record monitoring results
6. associate supporting evidence
7. identify findings or threshold events
8. initiate governance responses
9. link results to risks, controls, incidents, or changes
10. preserve monitoring history for assurance and improvement

The schema defines the structure of the governance information. It does not prescribe a particular monitoring platform, observability stack, database, workflow engine, or technical monitoring implementation.

​

Validation

Monitoring artifacts participate in the broader AIGO validation model. Relevant validation may include:

• schema validation
• reference validation
• traceability validation
• evidence relationships
• framework consistency
• document integrity
• repository health
• rule validation
• evaluation fixture validation
• evaluation orchestration validation

A passing validation result indicates that the tested artifact or repository content satisfies the implemented validation checks within the applicable scope. It does not establish legal compliance, certification, accreditation, or regulatory conformity.

​

Canonical source

Machine-readable schema: schemas/06-monitoring/06-AIGO-Monitoring-Schema-v0.1.json Schema documentation: schemas/06-monitoring/06-AIGO-Monitoring-Schema-Documentation-v0.1.md