The AI-Driven Operating Model: An Overview

An Operating Model is the structural blueprint that defines “who decides what, how, and at what speed” in order for an organization to execute its strategy. Redesigning the operating model is one of the most essential and difficult challenges in AI transformation.

What Is an Operating Model?

Definition

McKinsey (Rewired, 2023) defines the operating model as “the mechanism that converts strategy into executable organizational capabilities.” Specifically, it consists of five elements:[1]

Element	Content
Processes	Design of the workflows that generate value
Organizational structure	Arrangement of departments, teams, and roles
Governance	Decision-making authority and approval pathways
Technology	Systems and tools supporting operations
People & culture	The people and mindset that drive execution

Traditional operating models were designed around the principle of maximizing efficiency. With the arrival of AI, those very design principles are being questioned.

The Fundamental Impact of AI on Operating Models

AI brings three fundamental changes to the operating model.

graph LR
    A["Traditional Constraints"] -->|Removed by AI| B["Direction of Change"]
    A1["Decision cycles dependent on\nhuman processing speed"] --> B1["Real-time, continuous\ndecision-making"]
    A2["Work designed around\nexpert knowledge silos"] --> B2["Democratization of knowledge via AI\nand autonomous work execution"]
    A3["Fixed roles and\norganizational structures"] --> B3["Dynamic team composition and\nAI-human collaboration design"]

First, the speed and quality of decision-making changes. Areas that used to rely on weekly or monthly reporting cycles become real-time with AI. Second, the unit of work design changes. The starting point shifts from “can a human execute this?” to “what is the optimal division of labor between AI and humans?” Third, organizational boundaries change. Because AI agents can work across departments, the premise of siloed organizations breaks down.

Traditional vs. AI-First Operating Model

Comparative Analysis

Drawing on Accenture (Technology Vision 2024) and Deloitte (AI-Driven Operating Models, 2024) research, the key differences are:[3][4]

Dimension	Traditional Operating Model	AI-First Operating Model
Decision speed	Weekly–monthly cycles, hierarchical approval	Real-time–daily, data-driven autonomous judgment
Work design principle	Optimized for human execution capacity	Optimized for AI + human collaboration capacity
Role definition	Fixed job descriptions (JDs)	Dynamic skill portfolios with AI augmentation
Knowledge management	Tacit knowledge locked in individuals and departments	Organizational knowledge digitized and shared
Response to change	Year-end or periodic organizational changes	Continuous experimentation and adaptation
Value measurement	Cost reduction, productivity efficiency	Customer value creation, new revenue

Decision-Making Structure Comparison

graph TD
    subgraph "Traditional"
    T1["Frontline data collection"] --> T2["Report to manager"]
    T2 --> T3["Analysis and judgment at upper levels"]
    T3 --> T4["Downward communication of instructions"]
    T4 --> T5["Execution (weeks later)"]
    end

    subgraph "AI-First"
    A1["Real-time data"] --> A2["AI automatic analysis and anomaly detection"]
    A2 --> A3["Immediate alert to frontline employee"]
    A3 --> A4["Empowered frontline decision"]
    A4 --> A5["Execution (within hours)"]
    end

McKinsey’s “Agile Operating Model” and Its Relationship to AI

Overview of the Agile Operating Model

McKinsey’s Agile Operating Model replaces the traditional functional hierarchy with cross-functional teams (squads) pursuing customer-value-linked goals (OKRs). McKinsey (The Five Trademarks of Agile Organizations, 2018) finds that agile organizations rank in the 70th percentile on organizational health compared to non-agile organizations.[2]

graph TD
    subgraph "Agile Operating Model"
    V["Value Direction\n(North Star)"]
    V --> T1["Product Team A\n(Squad)"]
    V --> T2["Product Team B\n(Squad)"]
    V --> T3["Product Team C\n(Squad)"]
    T1 & T2 & T3 --> P["Platform\n(Shared Foundation)"]
    end

How AI Accelerates the Agile Operating Model

AI enhances the agile model in three ways:

1. Faster sprints AI-assisted code generation, testing, and documentation increases a squad’s output per sprint. GitHub research (2024) shows GitHub Copilot users experience an average 55% improvement in coding speed.[6]

2. Data-driven OKR management AI tracks and predicts OKR progress in real time, enabling squads to course-correct without delaying decisions.

3. Automated knowledge sharing AI-generated summaries and knowledge graphs dissolve the knowledge silos that form between squads.

Deloitte’s Human-Machine Collaboration Model

Model Concept

Deloitte (AI-Augmented Workforce: The Next Step in Human-Machine Collaboration, 2024) proposes a Human-Machine Collaboration Model that classifies types of work by AI suitability and designs the optimal collaboration pattern for each.[3]

quadrantChart
    title Human-Machine Collaboration: 4 Quadrants
    x-axis Low Autonomy --> High Autonomy
    y-axis Low Creativity & Judgment --> High Creativity & Judgment
    quadrant-1 Human-Led, AI-Supported
    quadrant-2 Fully Human
    quadrant-3 Fully Automated
    quadrant-4 AI-Led, Human-Supervised
    Data Entry Processing: [0.2, 0.15]
    Compliance Checks: [0.75, 0.2]
    Strategy Planning: [0.25, 0.85]
    Customer Service: [0.55, 0.65]
    Demand Forecasting: [0.8, 0.35]
    Creative Design: [0.3, 0.9]

Deloitte research shows that organizations that intentionally design Human-Machine Collaboration models see 23% higher employee engagement and 18% productivity improvement compared to those that do not.

Role Redefinition

In this model, the human role shifts from “executor” to three new roles:

New Role	Content	Required Skills
AI Orchestrator	Designs and supervises work by combining multiple AI agents	Prompt engineering, workflow design
Exception Handler	Judges complex or ethical situations AI cannot handle	Critical thinking, domain expertise
Relationship Builder	Trust-based relationships with customers and partners	Empathy, negotiation, cultural understanding

Three Core Elements of the AI-First Operating Model

Synthesizing frameworks from McKinsey, Deloitte, and BCG, three essential elements emerge that are common to AI-first operating models.[1][3][5]

graph TD
    OM["AI-First\nOperating Model"]
    OM --> S["Speed"]
    OM --> A["Adaptability"]
    OM --> V["Value Creation"]

    S --> S1["Shorter decision cycles"]
    S --> S2["Faster experimentation, learning, and improvement"]

    A --> A1["Continuous response to environmental change"]
    A --> A2["Dynamic updating of AI-human role division"]

    V --> V1["Continuous maximization of customer value"]
    V --> V2["Expansion toward new business models"]

Speed

Not just shortening decision cycles, but accelerating the entire cycle of experiment → failure → learning. Amazon’s “Two-Pizza Team” and API-first philosophy are representative examples of this design thinking.

Adaptability

The ability to continuously adapt to changes in the external environment (competitors, regulation, customer needs). Realized by AI detecting market signals in real time and automating micro-adjustments to operations.

Value Creation

Going beyond the inward-looking goal of efficiency to continuously creating new value for customers and society. In the AI-first operating model, data and customer insights feed directly back into product and service improvement.

FAQ

Q: Where should operating model transformation begin? A: McKinsey recommends starting with “domain identification.” Rather than company-wide transformation all at once, select a business domain where AI impact is high and data is available as a pilot, then scale the learnings.

Q: How do traditional functional organizations and agile organizations coexist? A: Deloitte research finds that a “bimodal” structure (stable functional organization + agile squads) is a more realistic transition path than full agile conversion. AI acts as the bridge connecting information between these two structures.

Q: What is the hardest transformation in the Human-Machine Collaboration model? A: Deloitte research identifies “role redefinition and employee mindset change” — not technology adoption — as the biggest barrier. Dispelling anxiety that AI will take jobs and teaching new collaboration patterns through experience is key.

References

1. McKinsey & Company, Rewired: The McKinsey Guide to Outcompeting in the Age of Digital and AI (2023)
2. McKinsey & Company, The Five Trademarks of Agile Organizations (2018)
3. Deloitte, State of Generative AI in the Enterprise (2024)
4. Accenture, Technology Vision 2024 (2024)
5. Boston Consulting Group, Winning with AI (2020)
6. GitHub, Does GitHub Copilot improve code quality? Here’s what the data says (2024)