1. Purpose and Position in the OS
MAT-01 is the central maturity assessment module in the P7 — Legal AI Maturity Mapping pillar. It provides a standardized, defensible way to measure how far a legal function has progressed in adopting and governing AI, and how sophisticated its deployed AI capabilities are.
The model is explicitly two-dimensional:
- Vertical axis — Adoption Stage (S1–S5): Organizational readiness, strategy, governance, implementation scope, and value realization.
- Horizontal axis — Augmentation Sophistication (L1–L4): Depth of AI capability and human–AI collaboration, from Advisor to Agentic Executor.
MAT-01 sits in the ecosystem as:
- GOV-01 → MAT-01 → STR-02 → USE-05 → MAT-01
- It also activates STR-03 and GOV-04 requirements for Level 4 (Agentic Tier) deployments.
The assessment is run at baseline and then quarterly, feeding the Annual Legal AI OS Index and Board-level reporting.
---
2. Risk Taxonomy 2026 Cross-Walk
MAT-01 embeds specific coverage of the Risk Taxonomy 2026 classes most relevant to legal AI:
- Class 1: Hallucination and accuracy — Assessed via Implementation Scope and Value Realization criteria, focusing on accuracy, validation rigor, and ROAI evidence.
- Class 3: Bias and fairness — Governance Maturity requires GOV-04 bias testing protocol compliance, especially for scaled and Agentic Tier deployments.
- Class 6: Shadow AI — Stage 1 explicitly surfaces shadow AI and defines remediation via AI BoM registration and governance activation.
- Class 7: Regulatory compliance drift — Governance Maturity checks regulatory monitoring and GOV-02 policy alignment.
- Class 9: Operational resilience — Level 4 (Executor) triggers STR-03 Class 9 risk modifier and mandatory Agentic Tier controls.
MAT-01 is therefore a cross-class risk instrument, with particular emphasis on Classes 1, 3, 6, 7, and 9.
---
3. Adoption Stage Maturity (Vertical Axis)
The Adoption Stage dimension measures how systematically the organization approaches AI. Stages are determined by a 100-point scorecard and mapped as follows:
- Stage 1 (0–19%): Exploring — Pilots in silos, no governance
- Isolated experiments, no AI governance, shadow AI prevalent.
- No AI BoM; vendor AI footprint unknown.
- Advancement requires: AI Task Force, GOV-01/GOV-02, STR-03 risk assessment, AI BoM initiation, basic VEN-02.
- Stage 2 (20–39%): Planning — Strategic alignment, prioritized use cases
- Documented AI strategy, active AI Task Force, prioritized use cases.
- Initial governance framework and GOV-02 approved use categories.
- Advancement requires: structured pilots, DAT-03 data governance, change management, ROAI evidence, integration capabilities.
- Stage 3 (40–59%): Implementing — Formal projects, early ROAI demonstrated
- Structured pilots with metrics, ROAI across multiple use cases.
- Formal change management, DAT-03 alignment, GOV-04 bias testing for all deployments.
- AI BoM current; VEN-03/VEN-04 on new vendors.
- Stage 4 (60–79%): Scaling — Enterprise-wide rollout and governance
- AI Center of Excellence in place; comprehensive governance applied consistently.
- Standardized vendor management with VEN-01 continuous scoring.
- GOV-03 Risk Register and DPS Defensibility Posture Statement operational.
- Stage 5 (80–100%): Realizing — AI embedded, continuous innovation
- AI fully embedded in appropriate workflows; continuous innovation culture.
- Advanced analytics and strategic vendor co-innovation.
- AI BoM integrated with continuous monitoring and quarterly reconciliation.
Governance Maturity is a core criterion within this axis and directly feeds the DPS Defensibility lens.
---
4. Augmentation Sophistication (Horizontal Axis)
The Augmentation Sophistication dimension measures the technical depth and human–AI collaboration model, independent of organizational maturity.
- Level 1: AI as Advisor — Provides insights and summaries
- Research summarization, document analysis, regulatory monitoring, case law analysis.
- High human involvement; low risk with strong human validation.
- Level 2: AI as Assistant — Automates discrete tasks
- Drafting assistance, data extraction, basic clause generation, matter intake routing.
- Medium human involvement; structured validation and exception handling required.
- Level 3: AI as Co-Creator — Collaborates on complex drafting
- Collaborative contract drafting, brief writing, complex review, regulatory filings.
- Medium human involvement; higher risk, requiring professional oversight and multi-level review.
- Level 4: AI as Executor (Agentic Tier) — Runs workflows under human oversight
- End-to-end workflow execution, constrained automated decisions, process orchestration, real-time compliance monitoring.
- Low human involvement; high risk and mandatory Agentic Tier governance.
Level classification is derived from a 100-point scorecard across Technical Capability, Human–AI Collaboration, Process Integration, and Risk Management.
---
5. Agentic Tier Governance Requirements (Level 4)
Any Level 4 deployment is designated Agentic Tier AI and must satisfy all of the following before deployment is authorized:
- Kill-switch mechanism — Documented emergency stop protocol with activation within 15 minutes; STR-07 notified on activation.
- Intervention logging — Full logging of AI decisions and exceptions; patterns recorded in GOV-03 Risk Register.
- Scope limitation verification — Clearly defined operational scope; any expansion requires STR-07 approval.
- Escalation protocol — Documented human escalation paths with response SLAs for all exception types.
- Continuous bias monitoring — GOV-04 continuous monitoring and monthly bias audits for all Level 4 workflows.
Failure on any control means the Level 4 deployment is not authorized.
---
6. Adoption Stage Assessment Methodology
Adoption Stage is scored on four equally weighted criteria (25 points each):
- Strategic Alignment (25%)
- Strategy alignment, executive sponsorship, resourcing, cross-functional collaboration, and regular strategic review.
- Governance Maturity (25%)
- GOV-01 framework, GOV-02 policy, STR-03 risk management, GOV-04 bias testing, AI BoM currency.
- Score 5: DPS operational, GOV-01 deployed, GOV-04 active, AI BoM current, STR-07 escalation, GOV-03 maintained.
- Score 3: Basic framework and GOV-02 in place.
- Implementation Scope (25%)
- Number and breadth of use cases, coverage, integration depth, adoption, and training effectiveness.
- Value Realization (25%)
- ROAI measurement across Protect, Comply, Grow, Transform; baselines, cost reduction, competitive advantage, and continuous optimization.
The total (0–100) maps to Stage 1–5 as defined above.
---
7. Augmentation Sophistication Assessment Methodology
Augmentation Sophistication is scored on four criteria:
- Technical Capability (30%)
- Technology sophistication, integration complexity, automation level, real-time processing, adaptive learning, and objective performance metrics.
- Human–AI Collaboration (25%)
- Oversight model, collaborative workflows, quality assurance, exception handling, and user experience.
- Process Integration (25%)
- Workflow automation depth, cross-system integration, end-to-end orchestration, monitoring, and continuous improvement.
- Risk Management (20%)
- Risk assessment, automated compliance monitoring, fail-safes, error recovery, and audit trail with explainability.
The total (0–100) maps to Levels 1–4:
- Level 1: 0–24%
- Level 2: 25–49%
- Level 3: 50–74%
- Level 4: 75–100%
---
8. DPS Lens Mapping
MAT-01 produces structured evidence for three DPS lenses:
- Adoption lens — Adoption Stage score (S1–S5) and quarterly progression.
- Sophistication lens — Augmentation Sophistication score (L1–L4) and oversight model.
- Defensibility lens — Governance Maturity sub-score (within Adoption Stage), including:
- AI BoM currency and coverage.
- GOV-04 bias testing status.
- STR-07 escalation log.
- GOV-03 Risk Register maintenance.
A Governance Maturity score below 3 indicates the function is below the minimum defensibility threshold for regulated and high-risk AI use. Advancement to Stage 3+ requires Governance Maturity ≥ 3.
---
9. Strategic Archetypes and Grid Positions
Grid positions are denoted as Stage × Level (e.g. S2×L2). Six archetypes guide interpretation:
- Governance-First — High Stage, Low Level: strong governance, limited AI depth; focus on sophistication.
- Technology-First — Low Stage, High Level: advanced AI, weak governance; DPS deficit requiring governance catch-up.
- Balanced Laggard — S1–S2 × L1–L2: early-stage on both axes; build foundations in parallel.
- Balanced Achiever — S3 × L3: mid-stage balance; scale and deepen simultaneously.
- Balanced Leader — S4–S5 × L3–L4: advanced balance; optimize, innovate, and shape standards.
- Agentic Early Adopter — Any Stage × L4: must satisfy Agentic Tier governance before sustaining Level 4.
The canonical progression is “up and to the right”. Stage-leading-Level indicates under-utilized governance; Level-leading-Stage indicates elevated STR-03 risk and DPS gaps.
---
10. AI BoM Integration and Maturity Gates
AI Bill of Materials (AI BoM) status is a key Governance Maturity indicator and is tied to stage gates:
- Stage 2 entry — AI BoM registry initiated; all current vendor AI systems catalogued.
- Stage 3 entry — AI BoM current; all new vendors complete VEN-03/VEN-04 and are recorded.
- Stage 4 entry — AI BoM includes sub-processors, risk classifications, and GOV-04 completion dates.
- Stage 5 sustaining — AI BoM integrated with continuous monitoring and quarterly reconciliation with contracts.
---