# 📐 Aetheris 7D AI Maturity Model — Full Rubrics & Indicators

This document contains the authoritative, detailed rubrics for the seven dimensions. Use it as the single source of truth for all scoring and analysis.

## Scoring Protocol

For each sub-dimension, collect evidence from at least three sources (interview, document, artifact/demo, survey). Score 1.0-5.0 in 0.5 increments. Document the primary evidence and any conflicting signals. The dimension score is the weighted average of its sub-dimensions (weights defined per engagement based on industry and strategy).

## Dimension 1: Strategic Leadership & Portfolio Management

Level 1.0: No enterprise AI strategy. Initiatives are entirely bottom-up or vendor-driven. Executive exposure is minimal or crisis-driven.
Level 2.0: AI mentioned in digital strategy. A few business units have local AI roadmaps. Executive sponsor exists but has limited bandwidth or authority.
Level 3.0: Board-approved AI strategy aligned to corporate goals. Formal prioritization framework and stage-gate process in use. Quarterly portfolio review established.
Level 4.0: AI portfolio dynamically managed with clear outcome ownership. Strategy refreshed based on realized value and external shifts. AI is a standing item in annual strategic planning.
Level 5.0: AI strategy directly shapes corporate and competitive strategy. The organization makes deliberate bets on AI to redefine markets. Strategic AI foresight function is mature.

Key Observable Indicators: existence and quality of AI strategy document; executive AI fluency demonstrated in conversations and decisions; percentage of AI spend with clear business case and owner; portfolio attrition rate (initiatives stopped); speed of strategic reprioritization.

## Dimension 2: Data Foundation for AI

Level 1.0: Critical AI-relevant data is fragmented, poorly documented, and of unknown quality. Every project starts with data discovery.
Level 2.0: Key datasets identified and partially governed. Data quality issues frequently derail AI projects. Feature stores or semantic layers are experimental.
Level 3.0: Enterprise data products and feature stores exist for priority domains. Lineage and quality SLAs are defined and monitored for AI workloads.
Level 4.0: Self-service, governed data products are the default for AI teams. Real-time and vector data capabilities are production-grade for approved use cases.
Level 5.0: Data is treated as a strategic product. Automated data quality and drift detection feed directly into model monitoring and strategic planning. Synthetic data generation is industrialized for privacy and augmentation.

## Dimension 3: AI Technology & Engineering Excellence

Level 1.0: Models built in notebooks and moved to production via manual handoff. No standardized evaluation, monitoring, or rollback.
Level 2.0: Individual teams have basic CI/CD for models. Monitoring is mostly performance metrics. LLM prompting is ad-hoc.
Level 3.0: Enterprise MLOps/LLMOps platform with standardized pipelines, model registry, and basic observability. Evaluation harnesses exist for major model classes.
Level 4.0: Full production-grade platform with automated drift detection, A/B and shadow testing, cost attribution, and governed prompt/version management. Red-teaming is routine for high-risk models.
Level 5.0: Platform enables safe, rapid experimentation and deployment at scale. Automated governance checks and continuous evaluation are embedded. Engineering velocity and reliability metrics are industry-leading.

## Dimension 4: Human Capital, Skills & Culture

Level 1.0: AI work performed by a handful of gifted individuals. High risk of key-person dependency. No formal AI literacy programs.
Level 2.0: Growing pockets of AI talent. Training is available but voluntary and uneven. Cultural tolerance for experimentation is localized.
Level 3.0: Defined AI career tracks and competency models. Mandatory AI literacy for relevant roles. AI translators exist in major business units.
Level 4.0: AI skills are systematically built and deployed. Performance systems reward responsible AI outcomes and cross-functional collaboration. Psychological safety for raising AI risks is high.
Level 5.0: The organization is a talent magnet for AI practitioners. AI-augmented decision making is part of every professional's expected capability. Failure is treated as valuable organizational learning.

## Dimension 5: AI-Enabled Process Architecture & Operating Model

Level 1.0: AI projects run in parallel to existing processes. No systematic redesign of decisions or workflows.
Level 2.0: Point solutions automate or augment narrow tasks. Human-AI handoff is poorly designed. Operating model experiments are isolated.
Level 3.0: Core processes have been redesigned with AI-augmented decision points. Clear human-AI teaming protocols exist. CoE provides standardized services and playbooks.
Level 4.0: AI is embedded in the primary value streams. Operating model choices (centralized, federated, hybrid) are intentional and regularly reviewed. Frontline adoption and override rates are measured and acted upon.
Level 5.0: The organization continuously re-architects processes around AI capabilities. Decision systems are instrumented for learning. The AI operating model itself evolves rapidly based on outcome data.

## Dimension 6: Governance, Risk, Compliance & Trust

Level 1.0: No AI-specific policies or oversight. Model risk is managed (or ignored) at the individual project level.
Level 2.0: Ad-hoc review processes for high-visibility models. Regulatory scanning is reactive. Shadow AI is widespread and unmeasured.
Level 3.0: Enterprise AI policy framework approved. Model inventory and basic risk classification exist. AI ethics or risk committee meets regularly.
Level 4.0: Integrated model risk management with automated controls for high-risk systems. Regulatory readiness (AI Act classification, conformity assessment) is proactive. Third-party AI risk is actively managed.
Level 5.0: Governance is largely predictive and embedded in platforms. The organization can demonstrate continuous compliance and can rapidly adapt to new regulations. Trust metrics are part of executive dashboards.

## Dimension 7: Value Realization & Continuous Improvement

Level 1.0: Success is declared by completion or anecdotal praise. No baseline or outcome tracking.
Level 2.0: Some projects track accuracy or user satisfaction. Economic value claims are mostly aspirational.
Level 3.0: Standardized value framework (cost, revenue, risk, experience) with baselines. Portfolio-level reporting exists.
Level 4.0: Value is attributed at the decision and process level. Realized vs projected ROI is tracked and used for portfolio decisions. Feedback loops from production to model improvement are closed.
Level 5.0: AI-driven decision quality and speed are primary management metrics. The organization measures and improves its own AI learning velocity. Value creation from AI is a sustained, compounding advantage.

## Cross-Dimension Patterns & Anti-Patterns

Documented in the living calibration notes (updated quarterly). Common patterns include: high tech maturity with low governance (typical of aggressive startups acquired by regulated firms); strong strategy with weak data (common in professional services); excellent pilots that never scale due to operating model and incentive misalignment.