Enterprise Coherence Model
Five coherence layers (Capital, Architecture, Operations, Risk & Compliance, Innovation), each with its metric and RAG threshold.
ECM™, PLCF™, VEGA™, TDG™ and CORE™, original constructions, and LEGATE™, an adaptation of the agent harness pattern. Together they govern the divestiture, depreciation and structural ageing of technologies and of delegations.
The integrating governance layer, above TOGAF, COBIT, SAFe and ISO 27001, that makes transformation outcomes measurable, board-reportable and causally traceable.
Large enterprises fail because they absorb change without structural discipline. Every acquisition, cloud migration and AI programme adds complexity: structural entropy. The dominant frameworks (TOGAF, ITIL, SAFe, ISO 27001, COBIT) each optimise their own domain and leave the coherence of the whole unmeasured. The standard reporting instruments of most CIOs do not give the board the structural visibility that material oversight of technology risk requires.
Five interdependent layers, each with a coherence metric, a governance owner, a board KPI and a RAG threshold calibrated for sector and strategic phase. An inter-layer causality model and four feedback loops, including F4, the AI Complexity Engine, connect the layers into a system. The scorecard is reported to the board quarterly and gives the five layer statuses without a composite score.
A five-layer scorecard the board can read, a causality analysis that identifies the primary intervention point, and a four-phase roadmap, from the Entropy Baseline Diagnostic (6 to 8 weeks) to the quarterly board reporting cycle.
The deliberate, governed closure of products: detect decay, run retirement as a go-to-market event, make the cost of retention visible.
Organisations have sophisticated disciplines for the inception, build and growth of products, and with rare exceptions no equivalent discipline for their closure. Technical debt compounds silently, applications in functional decay keep consuming support, licensing and compliance overhead, and the cost of maintenance crowds out investment in innovation.
Three interlocking instruments. The Product Decay Index (DXi) scores each product on five weighted dimensions and places it in the GREEN, AMBER, RED or BLACK band. Chargeback Mechanics activates at AMBER (DXi ≥ 3.6). The GTM Sunset Playbook opens at RED (DXi ≥ 6.0) and runs a four-phase sunset protocol, each phase with its exit criteria.
A quantified, repeatable end-of-life signal, a sunset programme that treats customers, data and regulatory obligations with the rigour of a launch, and a cost of debt made attributable to the business units that retain it.
Vendor portfolio design is an enterprise architecture decision.
Organisations manage vendor portfolios as commercial inventories. Their composition determines the integration topology, the security surface, the distribution of critical knowledge and the resilience of the technology estate, and none of these properties is visible to procurement governance. Vendor-embedded AI adds model dependency and exposure under the EU AI Act.
A four-level maturity model: Fragmented, Rationalised, Architected, Resilient. Four axes scored out of 100 points (Architecture Coverage 30, Governance Depth 25, Regulatory Alignment 25, AI & Sovereignty Compliance 20) produce the Integrated Vendor Score (IVS): below 40, Level 1; 85 and above, Level 4.
A board-reportable IVS that feeds CRS in ECM™ layer C1, and architecture artefacts (dependency map, concentration risk register, jurisdiction profiles, exit protocols) that DORA, NIS2 and the EU AI Act now require.
A framework for enterprise-scale arbitration: technical debt remediation treated as a capital allocation decision.
Technical debt has accumulated at enterprise scale for three decades without resolution: static analysis tools measure it, programme offices defer it, risk registers ignore it. Normalisation of deviance, the progressive redefinition of acceptable risk thresholds, sustains the accumulation, and incentives reward the addition of components and penalise their removal.
Three axes, Visibility, Arbitration and Regulation, produce three artefacts: the Systemic Debt Inventory (SDI), the Regime Assignment Register (RAR) and the Technical Risk Register (TRR). Debt is classified by three grains and four types: Foundational, Integration, Emerging Technology and AI-Generated Code. Each systemic item receives a regime, Remediate, Contain or Write-Off, or the Suspended Remediate state.
A Technical Risk Register for the audit committee in which each item carries a Principal Estimate, a Velocity Drag assessment, a Maturity Date and a named accountable owner: technical debt becomes a governed board commitment.
Governing technology competency as depreciating capital.
Most organisations treat technology competency as a stable resource whose value is assumed until it is unavoidably absent. It depreciates at market velocity, while its governance, owned by Human Resources, runs at annual rhythms. The risk register carries no entry for this exposure, and capital allocation decisions make no provision for it.
Three axes, Visibility, Arbitration and Regulation, produce the Competency Portfolio Inventory (CPI), the Treatment Assignment Register (TAR) and the Competency Risk Register (CRR). The taxonomy distinguishes three grains (individual, squad, portfolio) and one non-depreciating class, human judgment competencies. Each competency above the governance threshold receives a regime: Build, Source or Retire.
A Competency Risk Register for the board, alongside the financial and technical risk registers, that expresses exposure in capital terms: RAG status, quantified exposure and treatment owners.
Governing the delegated action of autonomous AI agents, and the ageing of that delegation. An adaptation of the agent harness pattern.
Autonomous AI agents now act inside enterprise systems: they pay, delete, send and deploy. The agent harness answers how an agent acts. It leaves three governance questions open: who authorised the delegation, who answers for its effects, and how long the authorisation remains valid. Delegation ages through permission creep, migration of oversight, model substitution and orphaned agents.
LEGATE™ takes up the agent harness pattern documented by LangChain, Anthropic, Microsoft, Salesforce and OpenAI and adapts it into a governance instrument, without adding any runtime component. Five components: Mandate, Boundary, Gates, Evidence, Interruption and Retirement. The unit of governance is the action class, ranked by autonomy level (A1 to A4, adapted from the AWS scoping matrix) and by reversibility class (R1 to R3, the framework's own construction).
A single board-reportable ratio, the Execution Legitimacy Score (ELS): the share of action classes in production that pass all five components, reported with five drift indicators. A single irreversible action class failing its human gate is enough to make the ELS Red.
Each framework addresses one dimension of technology governance. The documents listed below are the versions current in October 2026.
Five coherence layers (Capital, Architecture, Operations, Risk & Compliance, Innovation), each with its metric and RAG threshold.
Governance of product closure: Product Decay Index (DXi), GTM Sunset Playbook, chargeback of debt retention.
The vendor portfolio as an architectural artefact: four maturity levels, Integrated Vendor Score (IVS).
Technical debt arbitration: three grains, four types, the Remediate, Contain and Write-Off regimes, Technical Risk Register.
Technology competency as depreciating capital: the Build, Source and Retire regimes, Competency Risk Register.
An adaptation of the agent harness pattern: legitimacy of the action delegated to AI agents, measured by the Execution Legitimacy Score (ELS).
Two companion papers and the books that bring the corpus together.
The documents are available on request. Write to the author
The whitepaper of each framework is provided by the author on request.
Write to the author