Decompose the strategic thesis from Level 1 into a portfolio of Business Transformation Imperatives. The Methodology's central reframe: most organizations have AI use cases; they need Business Transformation Imperatives. Each Imperative is a domain-scoped, outcome-defined commitment with explicit resource allocation, a chartered domain owner, and the gating criteria for advancing through Levels 3 and 4. The engagement runs the C-suite through the imperative-definition methodology, surfaces the portfolio of imperatives the organization will pursue, allocates capital and talent to each, and stands up the monthly portfolio cadence that governs the work.

Establish the full governance architecture the Methodology specifies: risk classifications for AI deployments, accountability structures, oversight requirements, runtime enforcement specifications, board-level reporting integration, and the graduated autonomy spectrum from human-required to autonomous-with-oversight. Where most firms treat governance as a compliance overlay or a post-deployment audit function, the Methodology treats it as the operating system of the transformation that enables everything else. The governance built at Level 2 informs workflow design at Level 3, is enforced as runtime constraints at Level 4, and is continuously monitored at Level 5.

The Methodology's Communications track activates here. Communications begins at Level 2 because the workforce needs awareness and desire (ADKAR stages 1–2) early, well before training (ADKAR stages 3–4) at Level 4. The engagement designs the multi-tier communications cascade — C-suite through workforce — that prepares the organization for what's coming, addresses anxiety, builds buy-in, and creates the readiness conditions that the rest of the Methodology depends on. Operates organizationally agnostically: communications can sit in the CHRO function, an Enterprise PMO, the COO's organization, or a dedicated transformation office. Plaster Group works with whichever function holds responsibility.

A first-90-days engagement for newly chartered domain owners. Activates the charter from Level 2, builds the domain owner's AI fluency to the level required for Level 3 leadership, establishes the working cadence with the CAIO Department, and prepares the domain for the Capability Pathway and Workflow Redesign work that follows. Particularly valuable for the most under-served buyer in current Fortune 500: VPs and SVPs newly chartered as domain owners with no playbook for the role.

Decompose an imperative into the specific business capabilities required to deliver it. Capabilities are business abilities, not technology requirements — the Methodology's discipline is to think capability-first, technology-later. Produces the capability map that drives both the Workflow Redesigns at Level 3 and the Level 4 technology selection work that follows. This workstream precedes Workflow Redesigns.

Partnering with the CAIO's department, we deliver the fluency into the domains. Builds the substantive AI fluency required for ambitious workflow redesign — designer-side fluency for domain leadership, VPs, directors, senior managers, and business process analysts. Distinct from operator-side training at Level 4. Without this upstream investment, workflow redesigns under-use AI capabilities because the people designing them don't know what's possible. The Methodology's argument is that workflow redesign quality is gated by the design team's AI fluency; investing in this fluency upstream is what produces the ambitious workflow designs the cohort math requires.

The centerpiece of Level 3. A multi-month engagement applying the six-step workflow redesign methodology to a domain's business processes. The reframe is everything: rather than optimizing existing workflows around existing roles, the work redesigns workflows around what AI now makes possible — and designs people into the result deliberately. Includes the seven specific pitfall-prevention disciplines that catch the majority of redesign failures. When multiple domains are transforming in parallel, cross-domain coordination work is included here — the integration discipline that prevents the four standard breakpoints, operationalized through embedded translators and the cross-domain governance cadence. Every AI-enabled step is tagged with its underlying capability category, so each time a new or improved AI capability enters the market, the organization can immediately identify which specific workflow steps are candidates for improvement rather than reassessing every redesign enterprise-wide.

A diagnostic that surfaces, during workflow redesign, which data the redesigned workflows actually need, where it lives in the current state, and what the gaps are. Outputs feed directly into the Level 4 Data Architecture work. The Methodology's discipline: data readiness is discovered at Level 3 during workflow redesign, resolved at Level 4 in the data architecture build — not the other way around. Often runs late in or alongside Workflow Redesigns.

The Methodology's job redesign track — the second component of the parallel change-management work that runs from Level 2 (Communications) through Level 4 (Training). The engagement assesses the organizational impact of the redesigned workflows from L3.4, classifies roles into the three categories the Methodology identifies (augmented, consolidated, emergent), redesigns jobs accordingly, and produces the role-specific definitions that the Level 4 Training work will build competence against. Operates organizationally agnostically — sits with whichever function holds change management responsibility.

Workflow-driven evaluation methodology. The Methodology's discipline of evaluating AI technology against the specific workflow specifications produced at Level 3, not against generic capability demonstrations. Includes build/buy/partner decisions per workflow, composable architecture design (the Methodology's five-dimension framework), curated portfolio definition (the CAIO + CIO joint exercise), and vendor lock-in risk assessment.

AI integration is structurally different from any integration the organization has done before. The Methodology's five-difference diagnostic produces the architecture. The engagement covers the strategic integration path decision (incremental, comprehensive, or domain-based modernization), legacy system AI readiness assessment, orchestration layer design, agent identity and access architecture, and the data accuracy discipline embedded in pipelines.

The build, test, and deploy discipline calibrated specifically for AI's probabilistic, multi-environment, multi-vendor reality. Covers the eight engineering disciplines for configuration (memory and context management, tool integration, multi-agent coordination, model customization, evaluation and testing, governance and runtime guardrails, observability, cross-platform interoperability), AgentOps observability instrumentation wired during build (not retrofitted), the probabilistic testing methodology (multi-dimensional evaluation, continuous red teaming, golden datasets, behavioral fingerprinting, CI/CD-integrated evaluation, shift-right testing), the phased deployment architecture (the layered-cake model), the production monitoring discipline, the AI operations role architecture, environment management, and the iteration cycle operationalization (the four-option framework). Plaster Group will ensure the solutions these complexities require are carefully scoped and delivered for success.

The six new CDO capabilities the Methodology specifies: unstructured data integration pipelines, pipeline-embedded governance, the semantic layer (ontologies and knowledge graphs), continuous quality monitoring, AI training data standards, and AI data audit trail architecture (for EU AI Act high-risk compliance). Traditional data architecture served structured data to humans who provide their own context. AI consumes data differently — unstructured alongside structured, business context delivered with the data, real-time access vs. batch, governance traveling with the data rather than at boundaries. The CDO's organization must build six new capabilities that don't directly transfer from prior data work.

The Methodology's Training track — the third and final component of the parallel change-management work. Training builds Knowledge + Ability (ADKAR stages 3–4) and activates at Level 4 (after roles are defined at L3 and after technology is selected and built at L4.1–L4.3). Distinct from Enterprise AI Fluency at L3.3: that work targets designers; Training targets operators. The engagement covers the persona-based training architecture, the blended-model delivery (facilitator-led, hands-on practice, embedded reinforcement, continuous refreshment), the practice environments and simulations, the embedded coaching at point of need, and the three-level measurement framework (completion, application, outcomes). Operates organizationally agnostically.

The L5 → L1 feedback loop made operational. Inward sensing through the AI-Business Translators in the domains. Outward sensing through the AI Technology Strategists watching the capability environment. The CAIO's translation function turning signal into framed implications for the L1 triad. The cadence decision rhythm — monthly synthesis, quarterly thesis review, trigger-based escalation. The cascade mechanism that turns updated thesis back into refreshed imperatives. Stands alone for organizations already at Level 4 maturity, or as a continuation for clients Plaster Group has helped deliver through earlier levels.

Combines the diagnostic, build, and ongoing monitoring of the Methodology's five patterns of the self-optimizing organization (continuous, architectural, industrialized, learning-enabled, sensing-and-feeding-back). The setup phase applies the five-pattern diagnostic to surface which patterns are mature, which are absent, and which are masquerading as reporting layers rather than operating as actual patterns; then sequences the build to operationalize the missing patterns. The monitoring phase provides ongoing audit cadence — the bureaucracy-prevention discipline that catches the most common L5 failure mode (treating L5 as a reporting layer rather than an operating layer). For organizations already at Level 4 maturity or completing initial deployments, this is the engagement that operationalizes Level 5 as a structural shift rather than as intensified Level 3–4 execution.