Three Horizons Framework

Strategic Design Foundation

Three Horizons Framework provides the strategic architecture for Phased Internal Build by defining what the organization must accomplish at each horizon. Competitive analysis (Porter's) revealed we must defend core operations, build proprietary advantages, and create strategic options. Organizational constraints (7S) required a progressive, voluntary, bounded-risk approach. Three Horizons framework structures these requirements into three strategic objectives: H1 (Defend/Extend Core), H2 (Build Emerging Capabilities), H3 (Create Transformative Options). From these horizon objectives, we designed six phases to achieve the required transformation with exit optionality at every horizon boundary.

What Each Horizon Must Accomplish

Three Horizons Framework (McKinsey) provides the strategic architecture by defining three clear objectives: defend core operations, build emerging competitive advantages, and create transformative strategic options. Each horizon's objectives are derived directly from competitive analysis (Porter's), organizational constraints (7S), and stakeholder requirements.

Horizon 1: Defend and Extend the Core

Strategic Imperative: Core business under threat from lean competitors (Porter's), organization in crisis requiring immediate value delivery (7S, stakeholders), must defend by improving existing operations efficiency.

What H1 Must DeliverSuccess Criteria
Defend Core Business: Protect competitive position under threatCore operations measurably more efficient
Extend Core Efficiency: Improve operations without disruptionCompetitive vulnerability reduced
Preserve Client Value: No workflow changes, better serviceStaff self-initiating adoption (proof of value)
Generate Resources: ROI funds next horizonPositive ROI funding H2 consideration
Build Credibility: Success creates trust for H2 investmentDecision point: Can stop with defended/extended core

Horizon 2: Build Emerging Competitive Advantages

Strategic Imperative: Only defensible competitive advantage is proprietary capabilities on unique institutional knowledge (Porter's), must build with existing staff as AI talent is inaccessible (7S), stakeholders require non-commoditizable capabilities.

What H2 Must DeliverSuccess Criteria
Genuinely New Capabilities: Not faster existing work, but work we couldn't do beforeNew capabilities operational (not just efficiency gains)
Proprietary Advantage: Capabilities competitors cannot purchase or replicateStaff can do work previously impossible
Defensible Differentiation: Based on unique institutional knowledgeCompetitive differentiation achieved
Advanced Staff Capabilities: Internal expertise, not external dependencyProprietary advantage built on institutional knowledge
Competitive Moat: Sustainable advantage that strengthens over timeDecision point: Can stop with defended core + competitive advantages

Horizon 3: Create Transformative Strategic Options

Strategic Imperative: Rapidly changing AI environment requires strategic flexibility and ongoing innovation capacity (Porter's, 7S), stakeholders require strategic optionality beyond one-time transformation.

What H3 Must DeliverSuccess Criteria
Transformative Capabilities: Fundamentally different ways of operatingBreakthrough capabilities operational
Strategic Optionality: Options for future innovation and adaptationStrategic options created (not just current improvements)
Organizational-Scale Intelligence: Enterprise-wide AI capabilitiesInnovation capacity built (ongoing capability development)
Innovation Capacity: Internal expertise to continue evolvingOrganizational transformation complete
Future-Ready Organization: Sustainable competitive advantageSustainable advantage: Capability strengthens over time

From Horizons to Phase Design

Design Principle

Start with horizon objectives → Design phases to achieve them

Each horizon's strategic requirements drove the design of specific phases. Phases are not arbitrary divisions—they are purpose-built to accomplish horizon objectives while respecting organizational constraints.

Horizon 1: Phases 0, 1, 2 (Defend/Extend Core)

H1 Objective: Defend and extend core operations through existing capabilities augmentation

PhasePurpose & Design RationaleWhy H1?DeliverableValue
Phase 0: Registry Foundation (Month 0)Put registries in place to track experiments, data, and modelsMust establish tracking infrastructure before any AI work beginsExperiment registry, data registry, model registryEnables systematic AI development and governance
Phase 1: Knowledge Access (Months 1-3)Make trapped knowledge accessible to defend competitive positionWorks with existing knowledge, just makes it accessibleNatural language query, 4 languages, universal accessCore operations immediately more efficient
Phase 2: Task Automation (Months 4-9)Automate repetitive work in existing workflowsWorks with existing tasks and processes, just automates them50% repetitive work automatedExtended core efficiency, resources generated for H2

Why H1 Required Three Phases: To defend and extend core operations, we needed to establish tracking infrastructure (Phase 0), make existing knowledge accessible (Phase 1), and automate existing workflows (Phase 2). All three phases work with what already exists rather than creating new capabilities. Additionally, stakeholder requirements for quarterly ROI (S1) and exit optionality (S2) drove the decision to break H1 into three phases rather than one—each phase delivers independent value with decision points, enabling progressive investment with bounded risk.

H1 Investment: $106,000 ($0 Phase 0 + $62.4K Phase 1 + $43.6K Phase 2) | H1 Timeline: Months 0-9 | H1 Decision Point: Can stop with defended/extended core operations

Horizon 2: Phase 3 (Build Emerging)

H2 Objective: Build genuinely new competitive capabilities

PhasePurpose & Design RationaleWhy H2?DeliverableValue
Phase 3: Division Intelligence (Months 7-12)Create sophisticated analytical capabilities on unique institutional knowledgeCreates genuinely new capabilities that competitors cannot purchase or replicateAdvanced analysis, division-specific intelligence, sophisticated queriesStaff can do work previously impossible (not just faster)

Why H2 Required This Phase: To build genuinely new competitive advantages, we needed capabilities that don't exist today. Phase 3 was designed to create division-level intelligence—sophisticated analytical capabilities that enable work previously impossible, not just faster existing work.

H2 Investment: $31,200 (Phase 3) | H2 Timeline: Months 7-12 (overlap with H1 Months 7-9) | H2 Decision Point: Can stop with defended core + competitive advantages

Horizon 3: Phases 4 & 5 (Create Transformative Options)

H3 Objective: Create transformative strategic options

PhasePurpose & Design RationaleWhy H3?DeliverableValue
Phase 4: Agentic Discovery (Months 10-15)Enable cross-division experimentation in sandboxed environments to discover patterns and generate training data. H3 requires breakthrough capabilities—Phase 4 creates experimental foundation through siloed discoverySiloed experimentation generates training corpus for orchestrationExperimentation patterns, cross-division insights, training data corpusDiscovery of emergent organizational patterns
Phase 5: Orchestrated System (Months 13-18)Train orchestrator on Phase 4 experimental data to enable organizational-scale intelligence. Phase 5 transforms siloed discoveries into unified orchestrationTransforms how the organization operates (not incremental improvement)Trained orchestrator, integrated multi-agent system, organizational-scale intelligenceStrategic options for future innovation and adaptation

Why H3 Required Two Phases: To create transformative strategic options, we needed breakthrough capabilities beyond incremental improvements. Phase 4 generates training data through siloed experimentation across divisions—discovering patterns in sandboxed environments. Phase 5 uses this training data to build the orchestrator that unifies these discoveries into organizational-scale intelligence. Phase 4 creates the data that makes Phase 5 possible.

H3 Investment: $25,900 ($15.5K Phase 4 + $10.4K Phase 5) | H3 Timeline: Months 10-18 (overlap with H2 Months 10-12, Phase 4/5 overlap Months 13-15) | H3 Outcome: Full transformation complete, ongoing innovation capacity

Final Phase Design

Six phases designed from three horizon objectives:

PhaseTimelineHorizonPurposeDeliverableInvestment
Phase 0: Registry FoundationMonth 0H1Establish tracking infrastructure for systematic AI developmentExperiment, data, model registries$0 infrastructure
Phase 1: Knowledge AccessMonths 1-3H1Make existing knowledge accessible to defend competitive positionNatural language query, 4 languages, universal access$62,400
Phase 2: Task AutomationMonths 4-9H1Automate repetitive work in existing workflows to extend core efficiency50% repetitive work automated$43,600
Phase 3: Division IntelligenceMonths 7-12H2Create genuinely new analytical capabilities on institutional knowledgeAdvanced analysis, division-specific intelligence$31,200
Phase 4: Agentic DiscoveryMonths 10-15H3Enable cross-division experimentation to discover patterns and generate training dataExperimentation patterns, training data corpus$15,500
Phase 5: Orchestrated SystemMonths 13-18H3Train orchestrator on Phase 4 data for organizational-scale intelligenceTrained orchestrator, integrated multi-agent system$10,400

Total Investment: $163.1K Direct Investment ($11,100 infrastructure) | Total Timeline: 18 months | Decision Points: After Phase 2 (H1), Phase 3 (H2), Phase 4 (H3), or Phase 5 (H3)

Design Validation

The phase design accomplishes all horizon objectives:

Horizon ObjectivePhase(s) DesignedHow Phases Achieve Objective
H1: Defend CorePhases 0-2Registry foundation + knowledge access + task automation = defended/extended core
H2: Build EmergingPhase 3Division intelligence = genuinely new competitive capabilities
H3: Create TransformativePhases 4-5Agentic discovery (experimentation/training data) + orchestrated system = transformative strategic options

Strategic coherence: Every phase exists to accomplish a specific horizon objective. No phase is arbitrary. Phase 4 generates the training data that Phase 5 requires—they are sequential and interdependent within H3.