Organization Design for AI · Whitepaper
The HandsOn AI Operating Model
Six design domains. One Human-AI Interface at the core. The executive summary of our foundational whitepaper — the blueprint we use to redesign Organizations for the AI era.
6 min read
July 21, 2026
HandsOn Insights
Everyone is writing about how the real challenge of AI transformation lies in adapting the organization and its operating model to the new realities. At the same time, no one is offering a clear roadmap for how to systematically address this challenge. That’s exactly where the HandsOn AI Operating Model comes in. This article is the executive summary of our foundational white paper: The HandsOn AI Operating Model—Two Layers, Six Domains, One Interface.
Why an Operating Model Is Essential for the Effective Deployment of AI
McKinsey has found that AI high performers—that is, the nearly 6% of companies that attribute more than 5% of their EBIT to AI—are nearly three times more likely to have fundamentally redesigned their workflows. In their latest paper, “The Operating Model Advantage: Why AI Winners Are Rewiring Their Organizations,” they explore this topic in detail. When it comes to AI transformation, the central question is therefore whether an organization has structurally positioned itself appropriately for the use of AI. The choice of technology lies one level below and plays a relatively minor role in successful organizations.
The consulting market provides AI strategies, maturity models, and governance checklists. These describe where an organization stands today. What’s missing is the answer to the question of how an organization must be structurally reorganized so that AI can function within it. The HandsOn AI Operating Model closes precisely this gap. We have translated Galbraith’s Star Model, the center-led structures of Kates and Kesler, Warren’s axiomatic design, and the work of Jeroen van Bree into the AI-native context and used them to build a blueprint that helps organizations position themselves optimally for the age of AI.
The architecture: two layers, six domains, one core
The model organizes organizational work across two levels. The Foundation Layer clarifies what management must decide from an architectural perspective: strategy, structure, and system governance. The Activation Layer clarifies how work actually proceeds in the context of AI: decision architecture, process design, and capabilities. At the center of both layers lies the Human-AI Interface: it defines the collaboration between humans and AI systems.
D01 · Foundation
Strategy & Value Architecture
Where AI creates competitive advantage, where only efficiency. Use-case portfolio, build/buy/partner, cost of autonomy, in-house versus outsourced.
D02 · Foundation
Organizational Structure
Central CoE, federated model, or center-led hybrid with an AI Hub and Embedded AI Leads. Reporting lines, AI Owner and AI Steward roles.
D03 · Foundation
System Governance
Selection of the appropriate governance framework, risk tiering according to the EU AI Act, designated responsibilities per system, lifecycle governance.
D04 · Activation
Decision Architecture
The decision rights registry. For every decision type: which role decides at which autonomy level, and who can change that assignment. Escalation paths documented.
D05 · Activation
Process & Workflow Architecture
AI Overlay, AI-Integrated Redesign, AI-First Design — three approaches in parallel. Which process gets which treatment, in what sequence, at what investment.
D06 · Activation
Capabilities & Culture
Differentiated learning paths for five audiences from C-suite to frontline. The board decision on a high-risk system demands a different competency than daily copilot use.
At its core are four levels of autonomy—ranging from “Human-in-the-Loop,” where every AI recommendation is confirmed by a human, to “Human-in-the-Exception,” where AI operates autonomously and humans intervene only in borderline cases. Four design questions structure the approach for each decision type: Who decides? Who is liable? How does the system learn? Where are the limits?
In most organizations today, the human-AI interface exists unconsciously. It has evolved historically from individual tool decisions, approval workflows, and exception rules, but no one has a comprehensive view of the big picture.
Why We Deliberately Omit a Technology Domain
The model does not include a technology domain. Technology is a prerequisite for everything the model describes: D01 requires data infrastructure, D03 requires monitoring and logging, and D04 requires decision logging. We address these prerequisites in every project. They belong in procurement strategy, IT architecture, and platform strategy.
Specifically, this means: The question “Which LLM platform do we use?” belongs on the procurement agenda. The question “How do we manage the decisions this system makes on our behalf?” belongs on the organizational design agenda.
The Maturity Continuum — and the most dangerous position
The white paper introduces a maturity continuum ranging from Stage 0 (Unstructured) through Stage 1 (Augmented) and Stage 2 (Embedded) to Stage 3 (Agentic). An organization’s maturity level is determined by the configuration of all six domains. The HandsOn AI Operating Model uses the Multidimensional Maturity Profile across the six domains to determine maturity in detail.
The Transition Architecture
Chosing the right framework is only half the battle. Knowing how to move from one stage to the next is the other half. The Transition Architecture in the white paper describes, for the three critical transitions between Stages 0, 1, 2, and 3, which domains should be addressed first, which sequencing errors can derail the transformation, and which leadership decisions are embedded in each stage. This section is the operational core of the document and thus bridges the gap to implementation.
Three questions for your next board meeting
If your leadership team can answer three questions cleanly today, you are already in the transition from Stage 1 to Stage 2. If not, the root cause of your pilot-to-production gap sits in the operating model.
Q01 · Decision Architecture
01
Are your autonomy levels formally set?
For which decision types is it formally set at which autonomy level AI is allowed to operate, and who signed off on that assignment?
Q02 · System Governance
02
Do you know who owns each system?
Who is the named AI Owner and AI Steward for each system in production, and how often do we review that assignment?
Q03 · Process Architecture
03
Which core process was redesigned end-to-end?
Which core process has been redesigned end-to-end for AI, rather than getting AI as an overlay on an existing process?
Read the whitepaper. Book the diagnostic.
Six domains. One interface. A blueprint for the AI-native organization.
The whitepaper lays out all six domains in depth, the four autonomy levels of the Human-AI Interface, the Maturity Continuum, and the Transition Architecture. The Diagnostic is a four-to-six-week engagement that ends with a Multidimensional Maturity Profile, a gap analysis, and a prioritized design agenda for your leadership team.