Aims, Objectives & Stakeholder Map

This document is the programme spine. It maps each pillar to the stakeholders who own its findings and states the aim and objectives each pillar examines. Every objective carries an ID (e.g. P2-O1) so the two-tier interview questionnaire can link every question, sub-question, and probe back to a specific objective. It is aligned to the final Pillar 1–5 publications. Research design: this is a two-phase, mixed-method programme. Phase 1 (Pillars 1–5, complete) is secondary research — a structured synthesis of the documentary record, which establishes the conceptual framework, the aims and the objectives. Phase 2 (the field interviews, next) is primary research — semi-structured qualitative interviews with senior stakeholders, which test that framework in practice. All Phase 1 material is secondary data. Within it, sources are ranked by proximity to origin: the regulatory or legal text itself outranks a commentary on it, and a company’s own disclosure outranks a press report of it.

Programme thesis. The programme advances the proposition that enterprise AI success is determined not by which model an organisation uses, but by the architecture, operating model, economics, and governance it wraps around that model — the applied layer. Phase 1 sets out the evidence on which that proposition rests; Phase 2 tests it in the field. Pillar 1 establishes the thesis and the evidence; Pillars 2–5 each develop one load-bearing component for the stakeholder who owns it.

Stakeholder Map

Figure 1. Who owns each pillar. Each pillar is written for the senior stakeholder who must act on it; supporting roles are the secondary audience.

Task 1 · Stakeholder Mapping

Pillar	Primary stakeholders	Secondary / supporting	Focus
P1 · Beyond the Model	CEO, Board of Directors, Chief Strategy Officer (and all senior leaders)	Chief People Officer, Head of Innovation, Chief of Staff	Synthesis & strategy
P2 · Production AI Architecture	Chief Technology Officer (CTO), VP / Head of Engineering, Chief Architect, Head of Data & AI	CIO, Head of Product, Platform Engineering Leads, MLOps Leads	How systems are built
P3 · Operating Models	Chief Operating Officer (COO), Chief People Officer (CPO/CHRO), Head of Transformation	Business Unit Leaders, Head of Change Management, Strategy Director	How AI is delivered
P4 · Cost & Platform Landscape	CFO / Finance Director and FinOps leads who own the AI business case	CIO / CTO, Head of AI or Platform Engineering, procurement, and compliance	What it truly costs
P5 · Trust, Evaluation & Governance	General Counsel, Chief Risk Officer (CRO), Chief Compliance Officer — and the Head of AI Governance who must make the obligations operational	CISO, Head of ML / AI Platform Engineering, internal audit, and delivery leads	Risk & accountability

Detailed Stakeholder Profiles & Objectives

Each profile gives the primary and supporting stakeholders, the key idea, the aim, the objectives (with IDs and the interview theme each drives), and why the stakeholder needs the pillar.

PILLAR 1 · Beyond the Model Synthesis & strategy

Primary stakeholders. CEO, Board of Directors, Chief Strategy Officer (and all senior leaders)

Secondary / supporting. Chief People Officer, Head of Innovation, Chief of Staff

Key idea. A cross-cutting synthesis of what the first sustained period of enterprise AI in production has established, what it has recalibrated, and where the evidence points next — examining the widening distribution between organisations building AI capability and those treating it as a procurement category.

Aim. To establish what enables enterprise AI to create sustained production value, which early assumptions the public record has confirmed and which it has recalibrated — and to set the frame, the maturity baseline, and the forward research agenda for the five-pillar programme.

Obj.	Objective — what the pillar examines	Interview theme it drives
P1-O1	Examine the relationship between headline AI adoption and measured production impact, and test the widely cited (and contested) estimates of pilot-to-P&L conversion.	Strategy & leadership
P1-O2	Examine which structural conditions enable organisations to convert AI investment into capability — building capability rather than buying tools; line ownership alongside central expertise — and what is absent where progress slows.	Pilot vs production
P1-O3	Characterise the maturity ladder, examine where organisations locate themselves on it, and identify what characterises movement between rungs.	Pilot vs production / stratification
P1-O4	Establish the forward research agenda for the programme, and identify the questions practitioners most need answered.	Stratification & competitive awareness

Why this stakeholder needs this pillar. Every senior leader needs this as context for organisational AI investment decisions. It grounds strategy in evidence rather than expectation.

PILLAR 2 · Production AI Architecture How systems are built

Primary stakeholders. Chief Technology Officer (CTO), VP / Head of Engineering, Chief Architect, Head of Data & AI

Secondary / supporting. CIO, Head of Product, Platform Engineering Leads, MLOps Leads

Key idea. The evidence reviewed indicates that enterprise AI outcomes are shaped more by architecture than by model choice. The patterns wrapped around the model — retrieval, query rewriting, reranking, orchestration, evaluation — appear to account for most of the variation in system quality.

Aim. To examine how architecture shapes production AI outcomes, and to give technical leaders an evidence-based framework for moving from demo to reliable production.

Obj.	Objective — what the pillar examines	Interview theme it drives
P2-O1	Document the retrieval architecture patterns: naive RAG → hybrid → reranking → hierarchical/graph.	Data & retrieval
P2-O2	Map agentic maturity across a five-tier taxonomy (Tier 1: deterministic-with-LLM-glue → Tier 5: fully autonomous).	AI tools & automation
P2-O3	Examine which tiers are production-ready for which workloads, and what enables organisations to operate at higher tiers safely.	AI tools & automation
P2-O4	Provide a decision framework keyed to corpus characteristics, query complexity, latency budget, and error tolerance.	Infrastructure & integration

Why this stakeholder needs this pillar. CTOs and architects need this to move from demo to production. The evidence reviewed suggests the return on model upgrades is limited where the retrieval and architecture layers are not addressed first.

PILLAR 3 · Operating Models How AI is delivered

Primary stakeholders. Chief Operating Officer (COO), Chief People Officer (CPO/CHRO), Head of Transformation

Secondary / supporting. Business Unit Leaders, Head of Change Management, Strategy Director

Key idea. The evidence reviewed converges on the operating model as a stronger predictor of enterprise AI outcomes than technology choice. The same model deployed within two different operating structures appears to produce materially different results.

Aim. To examine how the operating model shapes enterprise AI outcomes, and to define the archetypes, components, and conditions that enable repeatable success across different organisational contexts.

Obj.	Objective — what the pillar examines	Interview theme it drives
P3-O1	Define the four operating-model archetypes: Centralised Platform, Centralised Delivery, Federated Platform, Federated Delivery/CoE.	Design authority & ownership
P3-O2	Examine the six recurring conditions of success — production reach, evaluation in production, systems-of-record integration, governance integrated with delivery, talent retention, executive sponsorship — and how far each is present in practice.	Across all P3 themes
P3-O3	Ground the analysis in documented enterprise programmes — including sustained deployments (JPMorgan Chase, Sanofi, Walmart) and publicly reported course corrections (Klarna, McDonald’s).	Design authority / funding & sponsorship
P3-O4	Examine the five components of an AI operating model — design authority, build capacity, governance regime, run model, funding flow — and which are load-bearing in practice.	Talent & capability; funding & sponsorship

Why this stakeholder needs this pillar. COOs own the operating model. Without it, AI is run as a technology project rather than an operational transformation. The publicly reported course corrections at Klarna and McDonald’s are instructive on that point.

PILLAR 4 · Cost & Platform Landscape What it truly costs

Primary stakeholders. CFO / Finance Director and FinOps leads who own the AI business case

Secondary / supporting. CIO / CTO, Head of AI or Platform Engineering, procurement, and compliance

Key idea. The dollar that matters most is rarely the per-token dollar. Model inference is only 20–40% of run-rate AI cost; retrieval, evaluation, observability, governance, and human review consume the rest — yet are invisible in headline pricing calculators.

Aim. To give enterprise decision-makers a fully-loaded view of what production AI actually costs, and a vendor-neutral basis for choosing platforms on the factors that genuinely determine fit — data residency, regulatory accreditation, existing stack, and language — rather than on headline model capability or per-token price.

Obj.	Objective — what the pillar examines	Interview theme it drives
P4-O1	Decompose the enterprise-AI cost stack across eight categories and test the proposition that inference is a minority of run-rate cost.	How organisations budget, meter, and track the full cost of AI in production.
P4-O2	Map the proportional cost shape across the five workload archetypes and the pilot / department / enterprise scale tiers.	Which workloads dominate spend, and which cost categories prove hardest to forecast.
P4-O3	Examine the global platform landscape — Western, Chinese, Indian, Korean, Japanese, Middle-Eastern — against capability, cost, data residency, and accreditation.	How platform shortlists are actually formed, and which evidence buyers trust.
P4-O4	Examine which factors govern platform selection in practice, and test the proposition that platform, identity, and regulatory gravity outweigh raw capability.	The decisive non-model factors in vendor selection and portfolio design.

Why this stakeholder needs this pillar. To build business cases on fully-loaded cost, anticipate the categories that overrun, and select platforms on the factors that actually decide fit — connecting the architecture of Pillar 2, the operating model of Pillar 3, and the governance of Pillar 5 to a defensible economic frame.

PILLAR 5 · Trust, Evaluation & Governance Risk & accountability

Primary stakeholders. General Counsel, Chief Risk Officer (CRO), Chief Compliance Officer — and the Head of AI Governance who must make the obligations operational

Secondary / supporting. CISO, Head of ML / AI Platform Engineering, internal audit, and delivery leads

Key idea. The pillar advances the proposition that evaluation and governance are a single operational system: governance without evaluation produces documents that are not enacted, and evaluation without governance produces dashboards no one acts on.

Aim. To examine evaluation and governance as a single operational system, and to give enterprises an inspectable maturity yardstick and a set of operational components and regulatory mappings that turn AI trust from aspiration into verifiable practice.

Obj.	Objective — what the pillar examines	Interview theme it drives
P5-O1	Examine the seven evaluation dimensions and five methods proposed as a production evaluation practice, and which are used in practice.	How organisations measure whether AI systems perform as intended once in production.
P5-O2	Examine the seven operational components of a working governance system, and the test that distinguishes governance which is operationally enacted from governance which is documented but not implemented.	What governance is actually implemented in delivery versus what is merely documented.
P5-O3	Map enterprise obligations across the EU AI Act, NIST AI RMF, and ISO/IEC 42001 to concrete engineering work items.	How regulatory obligations are translated into delivery, and what makes that translation work or slow.
P5-O4	Provide a combined maturity framework that locates a programme on a four-level ladder by inspection, and establish where enterprise programmes actually sit.	Where organisations sit on the evaluation–governance maturity ladder, and what enables or slows progress.

Why this stakeholder needs this pillar. To earn operational trust and meet binding obligations — EU AI Act high-risk duties from August 2026 — by ensuring documents are enacted and dashboards are acted on, closing the loop opened by the architecture of Pillar 2, the operating model of Pillar 3, and the economics of Pillar 4.

From aims to evidence. Each pillar’s objectives drive a two-tier field-interview instrument: a holistic session (~30 min) that sweeps all five pillars for breadth, and a granular session (~45 min) that goes deep on the respondent’s home pillar. Every question is mapped to the objective IDs above; the full instrument is in the accompanying Stakeholder Map & Questionnaire document.