Skip to content

Gartner's 2025 EA Diagnosis and What We Built For It

Context

Gartner's 2025 Leadership Vision for EA identified five systemic mistakes, summarized by Timo Elliott. The complementary 3 Trends article and EA Services Predictions webinar sharpen the message: EA must shift from centralized control function to federated, service-oriented, financially literate, AI-fluent capability.

This article maps each mistake to the Linked.Archi assets that address it — both the existing foundation and three extensions (Financial Architecture, AI Governance, EA-as-a-Service) built to close the remaining gaps.

The five mistakes and what addresses them

#1 — Outdated operating model

The problem: EA remains a centralized gatekeeper while organizations have gone federated and product-oriented.

What addresses it: The graph-based approach is structurally federated. Domain teams publish facts in their own notation (ArchiMate, C4, Backstage — whatever fits). SHACL shapes enforce shared principles without requiring central approval for every model change. Governance becomes collaboration, not control.

Key assets: arch:core as shared foundation, multi-language support, SHACL governance-as-code with sh:Warning by default (promoted to sh:Violation per local policy).

#2 — No portfolio modernization

The problem: Technical debt and bloated portfolios persist because rationalization requires structured, queryable knowledge that spreadsheets can't provide.

What addresses it: The TIME framework (timefw:) provides Tolerate/Invest/Migrate/Eliminate classification with evidence tracking. The Financial Architecture extension (fina:) attaches TCO models and migration cost estimates to applications. Together, you can query: "which Migrate-classified applications have the highest TCO and serve the fewest capabilities?"

Key assets: TIME framework, Financial Architecture extension, Reference Architecture (patterns/tactics), LeanIX interop.

#3 — Lack of financial acumen

The problem: EA can't articulate recommendations in ROI/TCO terms, weakening influence with executives.

What addresses it: The Financial Architecture extension (fina:) closes this gap:

  • fina:CostModel — TCO linked to any element
  • fina:CostBenefitAnalysis — structured CBA linked to ad:Option
  • fina:InvestmentCase — Run/Grow/Transform with projected ROI
  • fina:TechnicalDebtCost — makes invisible costs visible
  • fina:allocatedToCapability — cost-per-capability analysis

EA recommendations become: "modernize X because the 3-year TCO is EUR 1.25M, migration cost is EUR 400K, and NPV of Option A is EUR 320K at 8% discount rate."

See the Financial Architecture Practice Guide for worked examples.

#4 — Falling behind on AI

The problem: EA teams lack AI expertise and are sidelined from AI strategy.

What addresses it: Two complementary plays:

AI for EA — the RDF/OWL stack is natively machine-readable. The MCP server exposes the graph to AI agents for impact analysis, gap detection, and natural-language Q&A.

EA for AI — the AI Governance extension (aigov:) provides vocabulary for governing AI systems: EU AI Act risk classification, conformity assessments, bias assessments, human oversight plans, explainability reports. AI systems become architecture elements governed with the same rigor as everything else.

See the AI Governance Practice Guide.

#5 — No clear value proposition

The problem: EA is perceived as bureaucratic overhead with no measurable value.

What addresses it: The EA-as-a-Service extension (easvc:) models the EA practice itself as a service catalog:

  • Named services with SLEs and required capabilities
  • Engagement tracking from request through delivery
  • Stakeholder satisfaction scores
  • Maturity assessments (Gartner-aligned L1–L5)

This is the Internal Management Consultancy (IMC) model made concrete. "Here is the service, here is the stakeholder, here is the outcome, here is the satisfaction score."

See the EA-as-a-Service Practice Guide and the worked example at examples/ea-service/ea-service-example.ttl.

Beyond the five mistakes

Trend spotting

Gartner says EA must connect external change signals to architecture decisions. The decisions extension already supports this — ad:Trend is a first-class force type:

:GenAIAdoption a ad:Trend ;
    skos:prefLabel "Generative AI adoption in financial services"@en .

:ADR-AIStrategy a ad:Decision ;
    ad:influencedByForce :GenAIAdoption, :EUAIActEnforcement ;
    ad:justification "GenAI creates competitive pressure; EU AI Act creates compliance obligation."@en .

Trend-to-decision-to-roadmap traceability becomes queryable. What we don't provide is the sensing feeds — external market intelligence and regulatory monitoring must come from elsewhere.

Machine-mediated governance

Gartner says governance should be distributed and increasingly automated. This is already operational — SHACL + CI/CD is machine-mediated governance:

  1. Automated: SHACL shapes catch violations on every commit — no human needed for rule-based checks
  2. Human-on-the-loop: Validation reports surface issues for architect review
  3. Human-in-the-loop: Architecture Review Board handles judgment calls — trade-offs, exceptions, strategic direction

This escalation model maps to the aigov:OversightMode individuals (HumanInTheLoop, HumanOnTheLoop, HumanInCommand) — the same pattern applies to architecture governance generally.

End-to-end workflows

The extensions are building blocks. In practice, they compose into traceable workflows.

Portfolio modernization

1. Inventory & Classify
   ArchiMate elements + SHACL principle checks
   → Classify by lifecycle, criticality, ownership

2. Assess Fitness
   TIME Framework + evidence tracking
   → Score functional/technical/cost fit → assign disposition

3. Model Costs
   Financial Architecture
   → TCO models, migration estimates, cost-per-capability

4. Decide
   Architecture Decisions + CBA
   → Options with cost-benefit analysis, forces, rationale

5. Deliver & Track
   EA-as-a-Service
   → Engagement tracking, outcomes, satisfaction

Every step produces queryable data. At the end: "Which capabilities have the highest run cost?", "Which Migrate apps have the best ROI?", "How satisfied were stakeholders?"

Decision-centric governance

1. Issue Raised
   ad:Issue → link to affected elements, attach forces

2. Options Evaluated
   ad:Option → CBA per option, patterns/tactics, quality scenarios

3. Machine-Mediated Validation
   SHACL → principle checks, relationship validity, viewpoint conformance

4. Decision Recorded
   ad:Decision → selected option, justification, affected elements

5. Process Tracked
   ap:ArchitectureProcess → activities, roles, milestones
   easvc:ServiceEngagement → service delivery tracking

An auditor can trace from a deployed system back through the decision, the options considered, the forces, the governance process, and the EA engagement that delivered it.

References

Gartner sources