Enterprise AI Operating Model

The Enterprise AI Governance domain defines the policies, decision rights, and accountability within which the enterprise's use of Artificial Intelligence must operate. But policy and accountability do not exercise themselves. They require an organization—structures, roles, responsibilities, and ways of working through which the enterprise actually operates its AI. The Enterprise AI Operating Model provides this organizational dimension. It defines how the enterprise organizes itself to design, deliver, govern, and evolve Artificial Intelligence as a shared enterprise capability, and it is the subject of this domain.

An operating model describes how an organization is arranged to deliver a capability: the structures it establishes, the roles people play, the responsibilities and decision rights assigned to them, and the ways in which they work together. Applied to Enterprise AI, the operating model defines the organizational structures through which AI is delivered and operated, the roles and responsibilities of the people involved, the ownership and decision rights that determine who is accountable for what, and the collaboration models through which business and technology work together to produce AI that creates value. It is the human and organizational counterpart to the technical structures defined by the architecture and platform.

The relationship between the operating model and governance is close and complementary. Governance defines the policies, risk frameworks, and accountability that direct the enterprise's AI; the operating model defines the organizational structures through which governance is exercised and through which AI is actually delivered. Governance establishes that AI capabilities must have owners accountable for them; the operating model defines who those owners are and how they are organized. Governance establishes that decisions must be made within defined decision rights; the operating model establishes the roles and bodies that hold those rights. The two domains are inseparable: governance without an operating model is policy no one is organized to exercise, and an operating model without governance is organization without direction.

It is important to distinguish the operating model from an organizational chart. An operating model is not merely a diagram of reporting lines; it is a description of how the enterprise organizes to deliver a capability, encompassing structures, roles, responsibilities, decision rights, funding, and ways of working. Two organizations with identical reporting structures may operate Enterprise AI very differently depending on how they assign ownership, how they fund AI work, how business and technology collaborate, and how they organize the delivery and consumption of shared capabilities. The operating model addresses these dimensions of how the organization actually works, not merely how it is drawn.

The operating model is what allows Enterprise AI to be operated as a shared enterprise capability rather than as a series of disconnected projects. When AI is delivered as isolated initiatives, each assembles its own team, defines its own responsibilities, and works in its own way, and the enterprise accumulates organizational fragmentation to match its technical fragmentation. The operating model counteracts this by defining how the enterprise organizes to deliver AI consistently—establishing shared structures such as a platform team and a center of excellence, defining common roles and responsibilities, and organizing the relationship between those who provide shared capabilities and those who consume them. This organizational coherence is the counterpart to the technical coherence that the platform provides.

Within the broader Enterprise AI Operating Framework (EAIOF), the operating model occupies the position between governance and the domains that define how AI is built and run. The Reference Architecture and Platform provide the technical structures; governance provides the direction; the operating model provides the organization; and the domains that follow—Lifecycle Processes, the Engineering Framework, and Operations—define the processes and practices that this organization carries out. The operating model is thus the organizational foundation upon which the delivery and operation of Enterprise AI depend, giving the framework's processes and practices an organization to execute them.

This domain describes the Enterprise AI Operating Model from several complementary perspectives. It defines what an operating model is and why it matters for Enterprise AI. It establishes the principles that distinguish an effective operating model, the organizational structures through which AI is delivered, and the roles, responsibilities, decision rights, and ownership that give the organization its shape. It addresses the operation of the platform as a product, the ways of working and human–AI collaboration through which value is produced, the talent and organizational capability that Enterprise AI requires, and the funding and value management through which AI is sustained. Finally, it explains how the operating model serves as the engine of sustainable Enterprise AI.

For these reasons, the Enterprise AI Operating Model should be understood as the organizational dimension of Enterprise AI—the means by which the enterprise arranges its people, structures, and ways of working to deliver and operate AI as a shared capability. It gives governance an organization to exercise it, gives the platform an organization to provide and consume it, and gives the framework's processes an organization to carry them out. By defining how the enterprise organizes for AI, this domain enables Artificial Intelligence to be operated coherently, sustainably, and at scale, as a capability of the enterprise rather than a property of individual projects.

An Enterprise AI Operating Model is the definition of how an organization arranges itself to design, deliver, govern, and evolve Artificial Intelligence as a shared enterprise capability. It comprises the organizational structures through which AI work is organized, the roles that people play and the responsibilities they hold, the decision rights and ownership that determine who is accountable for what, the collaboration models through which business and technology work together, and the funding and ways of working through which AI is sustained. The operating model answers a question distinct from those addressed by the other domains: not what AI should do, how it should be structured, or how it should be governed, but how the organization arranges its people and work to make AI happen.

Within the Enterprise AI Operating Framework (EAIOF), the operating model is the organizational dimension that gives the rest of the framework an organization to execute it. The architecture and platform define technical structures; governance defines policies and accountability; the lifecycle and engineering domains define processes and practices. All of these require an organization to carry them out—people in defined roles, working within defined structures, holding defined responsibilities. The operating model provides this organization, translating the framework's technical structures, governance, and processes into an arrangement of people and work through which they are actually delivered.

The operating model is best understood through the dimensions it defines. Organizational structures describe how AI work is arranged across the enterprise—what central and distributed units exist, how the delivery of shared capabilities is organized, and how business and technology are brought together. Roles and responsibilities define the parts people play in the delivery and operation of AI and what each is accountable for. Decision rights and ownership determine who is authorized to make which decisions and who owns each capability and solution. Collaboration models define how the various parties work together, including how humans and AI systems themselves collaborate. And funding and ways of working define how AI work is resourced and how it is actually conducted. Together these dimensions constitute the operating model as a whole.

It is essential to distinguish the operating model from governance. The two are closely related but address different questions. Governance defines the policies, risk frameworks, decision criteria, and accountability that direct the enterprise's AI—the rules and the direction. The operating model defines the organizational structures, roles, and ways of working through which those rules are exercised and through which AI is delivered—the organization and the execution. Governance says that AI capabilities must have accountable owners and that decisions must be made within defined rights; the operating model establishes who those owners are, what roles exist, and how they are organized. Governance provides direction; the operating model provides the organization that acts on it.

It is equally important to distinguish the operating model from an organizational chart. A chart depicts reporting relationships; an operating model describes how the organization delivers a capability, which encompasses far more than reporting lines. The operating model addresses how ownership is assigned, how shared capabilities are delivered and consumed, how business and technology collaborate, how AI work is funded, and how the organization actually works day to day. Two enterprises with identical charts can have very different operating models, and it is the operating model, not the chart, that determines whether AI is delivered coherently. Reducing the operating model to a chart would omit precisely the dimensions that matter most.

A central purpose of the operating model is to enable Enterprise AI to be operated as a shared capability rather than as a collection of independent projects. This orientation shapes the entire operating model: it favors shared structures such as a platform team and a center of excellence over per-project duplication, common roles and responsibilities over ad hoc arrangements, and organized relationships between capability providers and consumers over isolated teams. The operating model is, in this sense, the organizational expression of the framework's central conviction that AI should be an enterprise capability, and it provides the organizational means by which that conviction is realized.

The operating model must also fit the specific character of the enterprise. There is no single correct operating model; the right structures, roles, and ways of working depend on the enterprise's size, its existing organization, its industry, its culture, and the maturity of its AI adoption. What the framework provides is not a prescribed organizational design but the dimensions an operating model must address and the principles that distinguish an effective one, from which each enterprise develops the operating model appropriate to its context. This adaptability is essential, because an operating model imposed without regard for the enterprise's character will not function regardless of its theoretical soundness.

Understood in this way, an Enterprise AI Operating Model is the organizational system through which the enterprise delivers and operates Artificial Intelligence as a shared capability. It defines the structures, roles, responsibilities, decision rights, collaboration, funding, and ways of working that give the framework's technical structures, governance, and processes an organization to execute them. It is distinct from governance, which it exercises, and from an organizational chart, which it exceeds, and it is shaped by the enterprise's own character. It is the means by which Enterprise AI becomes something the organization is arranged to do, rather than something that happens in scattered and disconnected ways.

The case for an Enterprise AI Operating Model rests on a recognition that organizations often learn slowly: scaling Artificial Intelligence is not primarily a technical challenge but an organizational one. An enterprise can acquire the best models, build a capable platform, and establish sound governance, and still fail to operate AI effectively if it has not organized itself to do so. The operating model matters because the obstacles to Enterprise AI at scale are, more often than not, organizational—unclear ownership, fragmented delivery, poor collaboration between business and technology, and the absence of the structures and roles that sustained AI requires. Addressing these obstacles is the purpose of the operating model.

The most fundamental reason the operating model matters is the transition from projects to enterprise capability. When AI is delivered as a series of independent projects, each assembles its own team, defines its own roles, and works in its own way, with the result that the enterprise's AI effort is organizationally fragmented even where its technology is shared. This fragmentation produces duplication of effort, inconsistency of practice, and an inability to accumulate organizational learning, because each project starts afresh. The operating model counteracts this by defining how the enterprise organizes to deliver AI as a shared capability, establishing the structures, roles, and ways of working that persist across initiatives. Without it, AI remains a project-level activity that cannot become an enterprise capability, however capable the underlying technology.

The operating model matters because ownership must be clear. Enterprise AI involves many capabilities, solutions, and decisions, and each requires an owner accountable for it. When ownership is unclear, capabilities go unmaintained, decisions go unmade, risks go unaddressed, and problems have no one answerable for them. Governance establishes that ownership must exist; the operating model establishes who owns what and how ownership is organized. This clarity of ownership is essential to operating AI reliably, because it is what ensures that every capability and solution has someone responsible for its quality, its governance, and its evolution. Ambiguous ownership is among the most common and damaging failures in scaling AI, and the operating model exists in large part to prevent it.

The operating model matters because shared capabilities require an organization to provide and consume them. The value of the Enterprise AI Platform depends on capabilities being provided once and consumed by many, but this arrangement does not organize itself. It requires a team accountable for providing the platform's capabilities as products, and it requires the consuming teams to be organized to consume them rather than to build their own. The operating model defines this relationship between capability providers and consumers, without which the platform's capabilities cannot achieve the reuse that justifies them. The organizational arrangement of provision and consumption is the counterpart to the technical arrangement of shared capabilities, and one cannot succeed without the other.

The operating model matters because business and technology must collaborate. Enterprise AI creates value only when technical capability is applied to business problems, which requires business and technology to work together closely throughout the delivery and operation of AI. When these are organized in isolation—technology building without business understanding, or business demanding without technical grounding—AI initiatives produce capable technology that solves the wrong problems or business ambitions that cannot be realized. The operating model defines the collaboration models through which business and technology work together, ensuring that AI is developed with business understanding and business value in view. This collaboration is not incidental to Enterprise AI but central to it, and organizing for it is a primary function of the operating model.

The operating model matters because Enterprise AI requires capabilities the enterprise may not yet possess. Delivering and operating AI at scale demands roles and skills that many organizations have not developed—in areas spanning AI engineering, data, governance, product management, and the operation of AI systems. Without an operating model that defines these roles and how the enterprise builds or sources the capability to fill them, the enterprise finds itself unable to staff its AI ambitions. The operating model addresses this by defining the roles Enterprise AI requires and organizing the development of the organizational capability to fill them, connecting to the broader work of adoption and enablement. Organizing for the human capability that AI requires is as important as organizing for its delivery.

The operating model matters, too, because how AI is funded shapes how it is delivered. Funding AI as a series of discrete projects produces short-lived initiatives that are built, delivered, and abandoned, whereas funding AI as an enduring capability supports the sustained investment that shared capabilities and long-lived solutions require. The operating model defines how AI work is funded and how its value is managed, and these choices profoundly affect whether the enterprise builds durable capability or a succession of disposable projects. Organizing the funding and value management of AI is therefore not a peripheral administrative concern but a central determinant of whether Enterprise AI can be sustained.

Finally, the operating model matters because it is the foundation of sustainability. The other reasons—clear ownership, shared provision and consumption, business–technology collaboration, organizational capability, and durable funding—all bear on whether the enterprise can sustain its AI over time rather than deliver a burst of initiatives that cannot be maintained. Sustainability is an organizational achievement: it depends on having an organization arranged to keep AI running, evolving, and creating value long after its initial delivery. The operating model provides this organization, and without it, even successful AI initiatives tend to decay because no organization is arranged to sustain them. This is the deepest reason the operating model matters: it is what allows Enterprise AI to endure as a capability rather than to flare and fade as a series of projects.

There is no single correct operating model for Enterprise AI, because the right organizational design depends on the character of each enterprise. But there are principles that distinguish operating models that work from those that do not—qualities that an effective operating model should exhibit regardless of the specific structures an enterprise adopts. These principles are not an organizational blueprint but design criteria, guiding how an enterprise shapes its operating model and providing a basis for assessing whether that model will support Enterprise AI effectively. They apply across all the dimensions of the operating model described in this domain.

The first principle is that the operating model should be capability-oriented rather than project-oriented. An operating model organized around discrete projects produces the fragmentation and impermanence that undermine Enterprise AI, whereas an operating model organized around enduring capabilities supports the shared provision, clear ownership, and sustained investment that Enterprise AI requires. Being capability-oriented means organizing around the durable capabilities the enterprise builds—the platform, its constituent capabilities, and the long-lived solutions that consume them—and treating projects as the means by which capabilities are created and evolved rather than as the fundamental unit of organization. This orientation is the organizational counterpart to the framework's treatment of AI as an enterprise capability.

The second principle is that the operating model should be federated, balancing central coordination with distributed delivery. Centralizing all AI work produces a bottleneck that cannot serve the whole enterprise and distances AI from the business problems it must solve, while fully distributing it produces fragmentation and prevents the accumulation of shared capability. An effective operating model federates: it establishes central structures for the shared capabilities, standards, and coordination that benefit from consolidation, while distributing delivery to the business domains where AI is applied. This federation mirrors the federated model of governance and is what allows the enterprise to combine consistency with responsiveness. The precise balance of central and distributed is a matter of enterprise context, but the principle of federation applies broadly.

The third principle is that the operating model should be product-oriented in how it organizes shared capabilities. The platform and its capabilities are most effective when provided as products, and this requires an operating model that organizes their provision accordingly—establishing teams accountable for capabilities as products, oriented around the teams that consume them, and responsible for their quality, evolution, and adoption. Product orientation contrasts with the project orientation that builds a capability and then disperses the team responsible for it, leaving the capability unowned. Organizing shared capabilities as products is what allows them to be provided, consumed, and sustained as the platform intends, and it is a defining characteristic of an effective operating model for Enterprise AI.

The fourth principle is that the operating model should ensure clear ownership and accountability. Every capability, solution, and decision within the enterprise's AI must have an identifiable owner accountable for it, and the operating model must assign this ownership unambiguously. Clear ownership is what allows AI to be operated reliably and governed effectively, because it ensures that responsibility for quality, risk, and evolution rests somewhere specific. This principle connects the operating model directly to governance, which requires the accountability that the operating model assigns, and it is among the most important properties of an effective operating model, because ambiguous ownership is among the most common causes of failure in scaling AI.

The fifth principle is that the operating model should promote business–technology alignment. Because Enterprise AI creates value only when technical capability meets business need, the operating model must be organized so that business and technology collaborate closely rather than operating in isolation. This means organizing delivery so that business understanding shapes what is built and technical grounding shapes what is promised, and establishing the collaboration through which the two work together throughout the delivery and operation of AI. An operating model that separates business and technology produces AI that is technically capable but misaligned with need, or business ambitions that cannot be realized; alignment is what ensures that AI creates value.

The sixth principle is that the operating model should be sustainable. Enterprise AI is not delivered once but operated and evolved over time, and the operating model must be arranged to sustain it—through durable funding, enduring ownership, ongoing capability development, and structures that persist beyond initial delivery. Sustainability requires that the organization be arranged to keep AI running and improving long after it is first built, rather than being organized only for delivery. This principle reflects the reality that the majority of an AI capability's life follows its initial creation, and that an operating model organized only for building will fail to sustain what it builds.

The seventh principle is that the operating model should be adaptable and evolvable. The enterprise's AI adoption matures over time, its needs change, and the field of Artificial Intelligence itself evolves, and an operating model that is fixed will become ill-suited to the enterprise it serves. An effective operating model is designed to evolve—adjusting its structures, roles, and ways of working as the enterprise's maturity grows and its circumstances change. Early-stage adoption may warrant a more centralized model that concentrates scarce capability, while greater maturity may warrant a more federated model that distributes it; the operating model should be able to make this transition deliberately rather than being locked into an arrangement suited only to one stage.

Taken together, these principles describe an operating model that is capability-oriented, federated, product-oriented, clear in ownership and accountability, aligned between business and technology, sustainable, and adaptable. An operating model exhibiting these qualities can organize the enterprise to deliver and operate Enterprise AI effectively, whatever specific structures it adopts, while an operating model that lacks them tends to produce the fragmentation, ambiguity, and impermanence that prevent AI from becoming an enterprise capability. The structures, roles, and practices described in the remainder of this domain are applications of these principles to the concrete organization of Enterprise AI.

The organizational structures of the operating model define how AI work is arranged across the enterprise—what units exist, what each is responsible for, and how they relate to one another. These structures are the scaffolding on which roles, responsibilities, and ways of working are hung, and the choices an enterprise makes about them shape how effectively it can deliver and operate AI. Because there is no single correct structure, this section describes the principal structural options and the considerations that guide the choice among them, rather than prescribing a particular arrangement.

The central structural question is how to balance centralization and distribution. At one extreme, an enterprise can centralize all AI work in a single unit that delivers AI for the whole organization. This concentrates scarce expertise, ensures consistency, and simplifies governance, but it distances AI from the business domains it must serve and becomes a bottleneck as demand grows. At the other extreme, an enterprise can distribute AI work entirely to its business domains, placing AI close to the problems it solves but fragmenting expertise, duplicating effort, and undermining consistency. Neither extreme serves Enterprise AI well at scale, which is why most effective structures fall between them, combining central and distributed elements in a federated arrangement.

The federated model, often realized as a hub-and-spoke structure, is the arrangement toward which most mature enterprises converge. In this model, a central hub provides the shared capabilities, standards, and coordination that benefit from consolidation, while distributed spokes deliver AI within the business domains where it is applied. The hub builds and operates the platform, establishes common practices, and provides expertise and support; the spokes apply these shared foundations to their specific business problems, close to the domain knowledge and business ownership that effective AI requires. This structure combines the consistency and efficiency of centralization with the responsiveness and business alignment of distribution, and it is the organizational counterpart to the federated model of governance.

Within the federated model, the platform team is a central structure of particular importance. Because the Enterprise AI Platform provides capabilities consumed across the enterprise, it requires a team accountable for building, operating, and evolving those capabilities as products. The platform team is the provider in the provider–consumer relationship that shared capabilities depend upon, responsible for the quality, reliability, and evolution of the capabilities that the rest of the enterprise consumes. Establishing a platform team with clear accountability for the platform is essential, because capabilities without an accountable provider decay and fail to achieve the reuse that justifies them. The platform team is where the technical shared capability of the enterprise is organizationally owned.

Another central structure common in Enterprise AI operating models is the center of excellence. Where the platform team provides shared technical capabilities, a center of excellence provides shared expertise, standards, and coordination—developing common practices, offering guidance and support to distributed teams, cultivating and disseminating expertise, and coordinating the enterprise's AI effort. The center of excellence is a mechanism for concentrating scarce knowledge and ensuring consistency without centralizing all delivery, allowing the enterprise to distribute AI work while still benefiting from shared standards and expertise. Its role is enabling and coordinating rather than delivering, which distinguishes it from a purely centralized delivery model.

The distributed elements of the structure—the spokes, or embedded delivery within business domains—are where AI is applied to business problems. These structures place AI delivery close to the domain knowledge, business ownership, and user understanding that effective AI requires, consuming the platform's shared capabilities and following the center of excellence's standards while focusing on the specific problems of their domain. The effectiveness of the distributed elements depends on their relationship with the central structures: they must be genuinely able to consume shared capabilities and receive support, rather than being left to build in isolation, which is why the provider–consumer relationship and the enabling role of the center of excellence are so important to the federated model.

The appropriate structure depends heavily on the enterprise's maturity and context. An enterprise early in its AI adoption, with scarce expertise and few initiatives, may benefit from a more centralized structure that concentrates its limited capability and establishes foundations. As adoption matures, expertise grows, and demand expands, a more federated structure becomes appropriate, distributing delivery while retaining shared foundations. The structure should therefore be expected to evolve as the enterprise matures, and the operating model should support this evolution rather than fixing a structure suited only to one stage. Matching the structure to the enterprise's maturity is one of the most consequential structural decisions, and getting it wrong—centralizing when distribution is needed, or distributing before foundations exist—is a common source of difficulty.

Whatever structure an enterprise adopts, its effectiveness depends on the clarity of relationships among its parts. The division of responsibility between central and distributed structures, the relationship between the platform team and its consumers, the role of the center of excellence in relation to delivery teams, and the connections between all of these must be clear, so that responsibilities do not fall between structures and work is not duplicated across them. Structural ambiguity produces the same failures as ambiguous ownership, and the value of a structure lies as much in the clarity of its relationships as in the units it establishes.

Understood in this way, the organizational structures of the operating model provide the arrangement of units through which the enterprise delivers and operates AI. A federated structure that combines a central platform team and center of excellence with distributed delivery in the business domains allows the enterprise to balance consistency with responsiveness, and to evolve as its maturity grows. These structures are the scaffolding on which the roles, responsibilities, and ways of working described in the remainder of this domain depend, and their design is among the most important choices the enterprise makes in organizing for Enterprise AI.

Organizational structures define the units through which AI work is arranged, but structures are populated by people playing defined roles. The roles and responsibilities of the operating model describe the parts people play in the delivery and operation of Enterprise AI and what each is accountable for. Defining these roles clearly is essential, because Enterprise AI brings together many kinds of expertise—business, product, architecture, engineering, data, governance, and operations—and without clearly defined roles, responsibilities overlap, gaps appear, and accountability becomes diffuse. This section describes the principal roles that Enterprise AI requires, understood as responsibilities to be assigned rather than as prescribed job titles.

It is important to treat roles as responsibilities rather than positions. A role is a coherent set of responsibilities that must be held by someone; how roles map to individuals, teams, and job titles depends on the enterprise's size and structure. In a small organization, one person may hold several roles; in a large one, a single role may be held by a team. What matters is that the responsibilities each role represents are held clearly by someone, not that the enterprise adopts a particular set of titles. The roles described here are therefore the responsibilities that Enterprise AI requires to be assigned, and the operating model's task is to ensure that each is held clearly and coherently.

Business and domain roles represent the responsibility for understanding the business problems that AI addresses and for owning the value that AI is meant to create. These roles bring the domain knowledge, business context, and value ownership without which AI solves the wrong problems or creates capability that delivers no value. They are responsible for articulating business needs, prioritizing AI work according to business value, and owning the outcomes that AI initiatives are meant to produce. Because Enterprise AI creates value only when applied to business problems, these roles are not peripheral to AI delivery but central to it, and their close involvement throughout is what keeps AI aligned with business need.

Product roles represent the responsibility for shaping AI capabilities and solutions as products that serve their users. Product roles bridge business and technology, translating business needs into a coherent vision for what should be built, prioritizing the work, and owning the ongoing evolution of a capability or solution as a product. This responsibility is especially important for the platform's shared capabilities, which must be managed as products serving the teams that consume them, and for AI solutions that serve business users. Product roles are what allow AI to be shaped around the needs of those it serves rather than driven purely by technical possibility, and they are central to the product orientation that the operating model requires.

Architecture roles represent the responsibility for the coherence and soundness of the enterprise's AI architecture. These roles ensure that AI capabilities and solutions are designed within the Reference Architecture, that they consume shared capabilities appropriately, that they fit coherently into the enterprise's AI ecosystem, and that architectural decisions are sound and consistent. Architecture roles connect the operating model to the architectural domains of the framework, and they are what ensure that the enterprise's growing collection of AI capabilities remains a coherent whole rather than fragmenting into inconsistent designs. Their responsibility spans both the design of individual solutions and the integrity of the architecture across the enterprise.

Engineering roles represent the responsibility for building and implementing AI capabilities and solutions. These roles encompass the specialized engineering that Enterprise AI requires—developing solutions by composing platform capabilities, building and evolving the platform's capabilities themselves, and applying the engineering practices that the framework's engineering domain defines. Engineering roles are where AI capabilities and solutions are actually constructed, and the expertise they require is among the capabilities that the enterprise must develop or source. Their responsibility is not merely to build but to build according to the standards and practices that ensure quality, consistency, and governability.

Data roles represent the responsibility for the knowledge and data on which AI depends. Because the quality and governance of information so directly determine the reliability of AI behavior, these roles—responsible for the enterprise knowledge, data, and the pipelines and structures that make information available to AI—are essential to effective Enterprise AI. Data roles ensure that the information AI consumes is available, of adequate quality, properly governed, and appropriately structured, connecting the operating model to the knowledge and data capabilities of the platform and to the enterprise's broader data governance. Their responsibility is foundational, because AI built on poorly managed information cannot be reliable regardless of the quality of its other elements.

Governance, risk, and compliance roles represent the responsibility for ensuring that AI is used within policy, that risk is managed, and that obligations are met. These roles—spanning governance, security, risk, legal, compliance, and ethics—bring the specialized judgment that AI governance requires and exercise the accountability that governance establishes. They connect the operating model directly to the governance domain, providing the people who set policy, assess risk, and ensure responsible and compliant use. Their responsibility is to ensure that the enterprise's AI remains aligned with its policies, tolerant of its risk appetite, and worthy of trust, and their involvement throughout delivery and operation is what makes governance operative rather than nominal.

Operations roles represent the responsibility for keeping AI capabilities and solutions running reliably once they are deployed. These roles ensure that AI systems remain available, performant, secure, and observable in operation, and they respond when problems arise. Operations roles connect the operating model to the operational domain of the framework, and they are responsible for the substantial portion of an AI capability's life that follows its deployment. Their responsibility is what ensures that AI, once built, continues to function reliably and to be governed and improved throughout its operational life, rather than being delivered and then neglected.

Understood together, these roles constitute the responsibilities that Enterprise AI requires to be assigned—business and domain, product, architecture, engineering, data, governance, and operations. The operating model's task is to ensure that each of these responsibilities is held clearly by someone within its structures, so that the full range of expertise that Enterprise AI demands is present and coordinated. How these roles are combined and mapped to individuals and teams depends on the enterprise, but the responsibilities themselves are what must be covered, and their clear assignment is what allows the enterprise to deliver and operate AI as a coherent capability rather than through an uncoordinated collection of efforts.

Structures and roles establish who is involved in Enterprise AI and what parts they play, but they do not by themselves determine who decides and who is accountable. Decision rights and ownership provide this. Decision rights define who is authorized to make which decisions; ownership defines who is accountable for each capability, solution, and asset. Together they give the operating model its accountability, ensuring that for every decision there is an authorized decision-maker and for every element of the enterprise's AI there is an accountable owner. Without clear decision rights and ownership, structures and roles produce activity without accountability, which is among the most common causes of failure in scaling AI.

Decision rights address a question that every organization must answer: who is authorized to make a given decision. Enterprise AI involves many decisions—whether an initiative proceeds, what capabilities it uses, how much autonomy it exercises, whether its risk is acceptable, how a shared capability evolves, and many more—and each must have a clear locus of authority. When decision rights are unclear, decisions are either not made, made by whoever happens to be present, or escalated unnecessarily, all of which impede effective delivery. Defining decision rights clearly ensures that decisions are made by the appropriate parties, at the appropriate level, with the appropriate authority, allowing the enterprise to act decisively while maintaining accountability for its choices.

Decision rights must be defined in a way that reflects the federated structure of the operating model. Some decisions belong at the enterprise level, where consistency requires that they be made once for the whole organization—decisions about enterprise policy, shared standards, and the direction of the platform. Other decisions belong with the delivery teams, close to the business problems and technical work they concern—decisions about how a particular solution is designed and built within established policy and standards. The allocation of decision rights across these levels is what allows the federated model to function: it enables most decisions to be made locally and quickly, while reserving for the enterprise level the decisions that genuinely require consistency. This allocation mirrors the distribution of authority in governance, with which decision rights are closely connected.

Decision rights are closely tied to risk, following the risk-based approach that governance establishes. The authority required for a decision should reflect the risk it carries: low-risk decisions can be made locally under delegated authority, while high-risk decisions require higher levels of authority and broader involvement. This connection between decision rights and risk ensures that the enterprise's attention is concentrated on the decisions that matter most, allowing routine, low-risk decisions to proceed without unnecessary escalation while ensuring that consequential decisions receive appropriate scrutiny. The mapping between risk and decision authority is one of the most important elements of the operating model's accountability, and it is shared with the governance model that classifies the risk on which it depends.

Ownership addresses accountability for the enterprise's AI assets—its capabilities, solutions, knowledge, and data. Every such asset must have an owner accountable for it: for its quality, its governance, its evolution, and its performance. Ownership is what ensures that the enterprise's AI does not accumulate unmaintained capabilities, ungoverned solutions, or neglected data, because for each of these there is someone answerable. The assignment of ownership is a primary function of the operating model, and it connects directly to governance, which establishes that ownership must exist, and to the architecture and platform, whose clear boundaries make it possible to assign ownership precisely. Because capabilities and solutions have well-defined boundaries, they can be owned as units, making ownership specific rather than diffuse.

The clarity of ownership depends on the alignment between ownership and the elements being owned. The architecture and platform define capabilities and solutions with clear boundaries, and ownership is most effective when it is assigned to these bounded elements—each capability owned by an accountable party, each solution owned by an accountable party. This alignment ensures that ownership covers the enterprise's AI completely, without gaps where no one is accountable or overlaps where accountability is contested. It also ensures that ownership is meaningful, because an owner accountable for a bounded capability can actually exercise responsibility for it, whereas ownership of something ill-defined cannot be exercised effectively. The correspondence between architectural boundaries and ownership is therefore central to the operating model's accountability.

Ownership must extend to shared capabilities, which present a particular challenge. Because the platform's capabilities are consumed by many teams, it might seem that no single party owns them, but the opposite must be true: shared capabilities require especially clear ownership, held by the platform team, precisely because so many depend on them. The owner of a shared capability is accountable to all its consumers for its quality, reliability, and evolution, which is a significant responsibility. Clear ownership of shared capabilities is what allows them to be relied upon, and its absence is what causes shared capabilities to decay when everyone consumes them but no one is accountable for them. The provider–consumer relationship of the platform depends on this clear ownership of the provided capabilities.

Decision rights and ownership together must form a coherent whole with no gaps or contradictions. For any decision, there should be a clear answer to who is authorized to make it; for any asset, a clear answer to who owns it; and the relationships between decision-makers and owners should be consistent. Ambiguity or contradiction in decision rights and ownership produces the paralysis, unaccountable action, and neglect that undermine Enterprise AI. The coherence of the enterprise's decision rights and ownership is therefore as important as their existence, and ensuring this coherence is a central task of the operating model, closely connected to the governance structures that depend on it.

Understood in this way, decision rights and ownership provide the accountability that gives the operating model its force. Decision rights ensure that decisions are made by authorized parties at appropriate levels, in proportion to risk and in keeping with the federated structure; ownership ensures that every capability, solution, and asset has an accountable owner, aligned with the boundaries the architecture and platform define. Together they ensure that the enterprise's AI is not merely staffed and structured but accountable—that for every decision there is a decision-maker and for every asset an owner. This accountability is what allows the operating model to support governance and to sustain the reliable delivery and operation of Enterprise AI.

The Enterprise AI Platform provides capabilities consumed across the enterprise, and the manner in which the enterprise organizes to provide and consume these capabilities determines whether the platform achieves the reuse and consistency that justify it. Operating the platform as a product is the organizational approach that makes this possible. It establishes the platform as a product with an accountable team, the consuming teams as its customers, and a service relationship between them, applying to the organization the product orientation that the platform domain establishes for the capabilities themselves. This section addresses how the operating model organizes the provision and consumption of shared capabilities.

The foundation of this approach is the platform team as product owner. Operating the platform as a product requires a team accountable for it—for building its capabilities, operating them reliably, evolving them over time, and ensuring that they serve the teams that consume them. This team owns the platform's capabilities as products, which means it is responsible not merely for their technical implementation but for their fitness for the consumers they serve, their reliability, their documentation, and their adoption. Establishing this accountable ownership is what allows the platform to be sustained and improved, and its absence is what causes shared capabilities to decay when they are built but not owned. The platform team is the organizational locus of the enterprise's shared technical capability.

The consuming teams are the platform's customers, and operating the platform as a product means treating them as such. The teams that build AI solutions consume the platform's capabilities, and their ability to do so easily and reliably determines the platform's value. Treating consuming teams as customers means understanding their needs, designing capabilities around those needs, making capabilities easy to discover and adopt, and treating the experience of consuming the platform as a first-class concern. This customer orientation is what distinguishes a platform that teams choose to use from one they seek to circumvent, and it is essential to the platform achieving the reuse that justifies it. A platform that is difficult to consume will be bypassed regardless of its technical quality, undermining the consistency and reuse it exists to provide.

The relationship between the platform team and its consumers is a service relationship, and the operating model must define it. This relationship encompasses how capabilities are made available, what consumers can expect of them, how support is provided, and how consumers' needs inform the platform's evolution. Defining this relationship clearly allows consuming teams to rely on the platform and to build upon it with confidence, knowing what they can expect and how to obtain support. An undefined or unreliable service relationship undermines consumption, because teams cannot build upon capabilities they cannot rely upon. The service relationship is the organizational counterpart to the technical interfaces through which capabilities are consumed, and both must be sound for the platform to succeed.

Operating the platform as a product implies a self-service orientation. Because a single platform team serves many consuming teams, the platform must be consumable without the platform team's individual involvement in every case, which requires that capabilities be discoverable, documented, and consumable through self-service. This self-service orientation is what allows the platform to scale to many consumers without the platform team becoming a bottleneck, and it is a defining characteristic of operating the platform as a product rather than as a service that must be individually provisioned. The organizational commitment to self-service complements the technical design of capabilities for self-service consumption, and both are necessary for the platform to serve the enterprise efficiently.

The platform's evolution must be driven by consumer needs and adoption. Operating the platform as a product means that its roadmap is shaped by the needs of the teams that consume it and that its success is measured by adoption and by the value it delivers to consumers. This orientation keeps the platform focused on serving the enterprise rather than on building capabilities for their own sake, and it provides the feedback through which the platform team learns which capabilities to build and how to improve them. A platform whose evolution is disconnected from its consumers tends to build capabilities that are not adopted, wasting effort and undermining the platform's value. Measuring the platform by adoption and consumer value is what keeps it aligned with the enterprise it serves.

Operating the platform as a product also requires attention to adoption and enablement. Providing a capability is not sufficient; consuming teams must be enabled to adopt it, through documentation, guidance, support, and the cultivation of the skills required to use it. This enablement is part of operating the platform as a product, because a product that customers cannot adopt delivers no value, and it connects the operating model to the broader work of adoption and enablement addressed elsewhere in the framework. The platform team's responsibility extends beyond building capabilities to ensuring that they are adopted, which requires investment in the enablement of the teams that consume them.

The product orientation must be sustained through durable funding and ownership, which connect operating the platform as a product to the funding concerns of the operating model. A platform operated as a product requires enduring investment, because it is built and evolved continuously rather than delivered once, and it requires enduring ownership, because it must be sustained over time. Funding the platform as a discrete project rather than an enduring product undermines its provision, because it leaves the platform unowned and unmaintained once the project ends. The organizational commitment to operating the platform as a product therefore depends on a corresponding commitment to funding and owning it durably.

Understood in this way, operating the platform as a product is the organizational approach through which the enterprise provides and consumes its shared AI capabilities. By establishing the platform team as accountable product owner, treating consuming teams as customers, defining a clear service relationship, orienting toward self-service, driving evolution by consumer needs, investing in adoption, and sustaining the platform through durable funding and ownership, the operating model allows the platform to achieve the reuse and consistency that justify it. This organizational approach is the counterpart to the technical product orientation of the capabilities themselves, and together they are what allow the Enterprise AI Platform to serve the enterprise as intended.

Structures, roles, and ownership define how the enterprise is arranged for Enterprise AI, but they do not describe how the people within that arrangement actually work. Ways of working address this: the practices through which teams collaborate, the manner in which business and technology come together, and—distinctively for Enterprise AI—the manner in which humans and AI systems themselves collaborate. How the enterprise works is as consequential as how it is structured, because sound structures populated by poor ways of working produce poor outcomes. This section addresses the collaboration through which Enterprise AI is actually delivered and operated, including the collaboration between people and the intelligent systems they build.

The foundation of effective ways of working in Enterprise AI is collaboration between business and technology. Because AI creates value only when technical capability is applied to business problems, business and technology must work together closely and continuously, not in sequence or in isolation. This means that business understanding shapes what is built from the outset, that technical grounding informs what is promised, and that the two collaborate throughout the delivery and operation of AI rather than meeting only at the boundaries. Ways of working that separate business and technology—handing requirements from one to the other without ongoing collaboration—produce AI that is technically capable but misaligned with need, or business ambitions that cannot be realized. Continuous business–technology collaboration is the practice on which value-creating AI depends.

Enterprise AI is best delivered through cross-functional collaboration that brings together the range of expertise it requires. Because delivering AI draws on business, product, architecture, engineering, data, governance, and operations expertise, effective ways of working bring these together around the work rather than dividing the work among them sequentially. Cross-functional collaboration allows the different kinds of expertise to inform one another throughout delivery, so that architectural, governance, and operational considerations shape a solution as it is built rather than being applied after the fact. This integrated way of working is what allows the many concerns that Enterprise AI raises to be addressed coherently, and it reflects the embedded, by-design approach that governance and engineering both favor.

Ways of working must also be iterative and adaptive, reflecting the nature of AI. Because AI behavior is probabilistic and emerges from experimentation, and because the understanding of what an AI capability should do develops as it is built, effective ways of working for AI are iterative—building, evaluating, learning, and refining—rather than attempting to specify everything in advance. This iterative character connects ways of working to the lifecycle and engineering domains of the framework, which define the processes and practices through which AI is developed. Ways of working that demand complete specification before development, as though AI were deterministic software, are ill-suited to the experimentation that effective AI development requires.

Distinctive to Enterprise AI is the concern with human–AI collaboration itself. As AI capabilities become more capable and more autonomous, the manner in which people and AI systems work together becomes a central concern of the operating model, because much of the enterprise's work will increasingly be performed through collaboration between humans and intelligent systems. This collaboration is not merely a matter of using AI tools; it concerns how work is divided between people and AI, how people direct and oversee AI, how AI augments human judgment, and how the two combine their respective strengths. Designing this collaboration deliberately is a responsibility of the operating model, because the effectiveness of the enterprise increasingly depends on how well its people and its AI work together.

Effective human–AI collaboration rests on a complementary division of work between people and AI. People and AI systems have different strengths: AI can process information, generate content, and act at scale and speed, while people bring judgment, context, values, and accountability. Effective collaboration assigns to each what it does best—allowing AI to handle what it handles well while reserving for people the judgment, oversight, and accountability that require human involvement. This complementary division is the practical expression of the human-centered principle that runs throughout the framework, and it is what allows the enterprise to capture the value of AI while preserving the human judgment and accountability that consequential work requires. Determining this division deliberately, rather than allowing it to form by default, is a key concern of the operating model.

Human–AI collaboration also depends on appropriate human oversight and control, connecting ways of working to the governance of autonomy. As people work alongside increasingly autonomous AI, the ways of working must preserve meaningful human oversight where the stakes require it, ensuring that people retain the ability to direct, review, and intervene in AI behavior. This oversight must be designed to be meaningful rather than nominal, giving people the information, time, and authority to exercise genuine judgment. The design of human–AI collaboration is therefore closely connected to the governance of autonomy and oversight, and it is where those governance requirements are realized in the actual working practices of the enterprise. Ways of working that grant AI autonomy without preserving meaningful oversight fail this requirement, however effective they may appear.

The ways of working must be supported and cultivated rather than assumed. Effective collaboration—between business and technology, across functions, iteratively, and between humans and AI—does not arise automatically; it must be supported through the practices, tools, and culture that the enterprise cultivates. This cultivation connects the operating model to the work of adoption and enablement, which develops the culture and capabilities that effective ways of working require. The operating model establishes the collaboration that Enterprise AI requires; the cultivation of that collaboration is a shared concern with the enablement of the enterprise's people, addressed further in the adoption domain of the framework.

Understood in this way, ways of working and human–AI collaboration describe how the people of the enterprise actually work to deliver and operate AI, and how they collaborate with the intelligent systems they build. Continuous business–technology collaboration, cross-functional and iterative practices, and the deliberate design of human–AI collaboration with meaningful oversight together determine whether the enterprise's structures and roles produce effective outcomes. As AI becomes more capable and more woven into the enterprise's work, the design of these ways of working—especially the collaboration between people and AI—becomes an increasingly central concern of the operating model, because the effectiveness of the enterprise increasingly depends upon it.

An operating model defines the structures, roles, and ways of working through which the enterprise delivers AI, but these require people with the skills to fill them. Talent, skills, and organizational capability address the human capability that Enterprise AI demands: the skills the roles require, how the enterprise builds and sources them, and how it develops the organizational capability to deliver and operate AI over time. This is a concern of the operating model because an operating model whose roles cannot be filled is an organizational design that exists only on paper. The deeper work of enabling the broader workforce and managing organizational change belongs to the adoption and enablement domain; this section addresses talent and capability as the operating model's own concern with staffing and sustaining its structures.

Enterprise AI requires a range of specialized skills that many organizations have not fully developed. The roles the operating model defines—spanning business, product, architecture, engineering, data, governance, and operations—each demand particular expertise, and some of this expertise is scarce and newly important. AI engineering, the management of AI products, the governance of AI risk, the management of the knowledge and data that AI consumes, and the operation of AI systems all require skills that are in high demand and short supply. The operating model must recognize these skill requirements explicitly, because the enterprise cannot deliver AI through roles it cannot staff, and the scarcity of AI skills is among the most common constraints on scaling Enterprise AI.

The enterprise builds its AI capability through a combination of developing, sourcing, and partnering. It can develop capability by building the skills of its existing people, sourcing it by hiring those who possess the needed skills, and partnering by drawing on external expertise where appropriate. Each approach has its place: developing existing people builds durable capability and leverages institutional knowledge but takes time; hiring brings capability quickly but competes in a scarce market; partnering provides access to expertise the enterprise lacks but does not build lasting internal capability. The operating model must determine how the enterprise combines these approaches to fill its roles, recognizing that reliance on any one alone is usually insufficient, and that the right combination depends on the enterprise's circumstances and the scarcity of the skills in question.

Developing existing people is central to building durable organizational capability. While hiring and partnering address immediate needs, the enterprise's lasting capability comes from developing the skills of its people, which builds capability that endures and that is grounded in the enterprise's own context and knowledge. This development is not confined to technical specialists; it extends to the broader set of people whose roles are changing as AI becomes part of the enterprise's work, from business people who must understand what AI can do to leaders who must govern it. The operating model's concern with developing capability connects it to the adoption and enablement domain, which addresses the broader development of the workforce; the two share the goal of building the human capability that Enterprise AI requires.

The concentration and distribution of scarce skills is a structural concern that connects talent to the organizational structures of the operating model. When AI skills are scarce, concentrating them—in a central platform team and center of excellence—allows the enterprise to make the most of limited expertise and to build depth, while distributing them places capability close to the business problems it serves but spreads scarce expertise thin. The appropriate balance depends on the enterprise's maturity: early adoption often warrants concentration, while greater maturity and a larger pool of skills allow more distribution. This connection between talent and structure is why the operating model's structural choices and its talent strategy must be made together, since a structure the enterprise cannot staff will not function, and scarce talent poorly distributed will not deliver.

Building organizational capability is a sustained effort rather than a one-time acquisition. The skills Enterprise AI requires evolve as the field advances, the enterprise's needs change as its adoption matures, and the people who hold critical skills move on. The operating model must therefore treat the building of capability as an ongoing responsibility—continuously developing skills, refreshing them as the field changes, and sustaining the capability of the organization over time. Capability treated as a one-time acquisition decays, because the skills that were sufficient become outdated and the people who held them depart. This sustained character of capability building reflects the broader sustainability principle of the operating model and the continuous-evolution disposition of the framework as a whole.

The development of capability must keep pace with the enterprise's AI ambitions. An enterprise's AI strategy defines what it intends to achieve, and its capability must be sufficient to deliver those ambitions; a mismatch between ambition and capability produces either unrealized strategy or overreach that fails. The operating model must therefore ensure that the enterprise's capability development is aligned with its ambitions—building the skills required to deliver what the enterprise intends, and tempering ambitions that outrun the capability available. This alignment connects the operating model to the enterprise's strategy, ensuring that the human capability the enterprise builds is matched to what it intends to accomplish with AI.

Understood in this way, talent, skills, and organizational capability address the human capability without which the operating model's structures and roles cannot function. By recognizing the specialized skills that Enterprise AI requires, combining the development, sourcing, and partnering through which capability is built, concentrating and distributing scarce skills in keeping with the enterprise's structure and maturity, sustaining capability over time, and aligning it with the enterprise's ambitions, the operating model ensures that its organizational design can actually be staffed and sustained. This concern connects closely to the adoption and enablement domain, which addresses the broader development of the workforce, and together they ensure that Enterprise AI has the people it requires to be delivered and operated as an enterprise capability.

How the enterprise funds its Artificial Intelligence, and how it manages the value that AI produces, profoundly shapes what the enterprise is able to build. Funding determines whether AI is sustained as an enduring capability or delivered as a succession of disposable projects; value management determines whether the enterprise directs its AI investment toward what creates value and can demonstrate the returns it earns. These are concerns of the operating model because funding and value are organizational choices that determine the character of the enterprise's AI, not merely financial administration. This section addresses how the enterprise funds and sustains its AI, and how it manages the value that AI is meant to create.

The most consequential funding choice is between funding projects and funding capabilities. Traditional funding treats work as discrete projects, each funded to deliver a defined output and then concluded. Applied to Enterprise AI, project funding produces initiatives that are built, delivered, and then abandoned, because no funding sustains them once the project ends—an arrangement fundamentally at odds with the shared, enduring capabilities that Enterprise AI requires. Funding capabilities, by contrast, provides enduring investment in the capabilities the enterprise builds—the platform, its constituent capabilities, and long-lived solutions—supporting their continuous development, operation, and evolution. The shift from project funding to capability funding is among the most important changes the operating model must effect, because it determines whether the enterprise builds durable capability or a series of impermanent projects.

Capability funding is essential to the platform in particular. The Enterprise AI Platform is built and evolved continuously and must be operated and sustained over time, which requires enduring investment that project funding cannot provide. A platform funded as a project is built and then left unmaintained once the project concludes, undermining the very reuse and reliability that justify it. Funding the platform as an enduring capability, with sustained investment in its provision, operation, and evolution, is what allows it to be operated as a product and to serve the enterprise reliably. The funding model and the product operating model of the platform are therefore closely connected, and both depend on the commitment to fund the platform durably rather than as a discrete initiative.

Funding must be allocated according to value and strategy. The enterprise's AI investment is finite, and the operating model must ensure that it is directed toward the capabilities and solutions that create the most value and best serve the enterprise's strategy. This requires a means of prioritizing AI investment—understanding the value that different capabilities and solutions offer, weighing this against their cost and risk, and allocating funding accordingly. This prioritization connects funding to the business and product roles that own value, and to the enterprise's strategy, which defines what the enterprise intends to achieve. Funding allocated without regard for value and strategy tends to be dispersed across initiatives that do not serve the enterprise's priorities, wasting scarce investment.

Value management addresses whether the enterprise's AI actually creates the value it is meant to create. It is not sufficient to fund AI; the enterprise must manage the value that funding is meant to produce—defining the value each initiative is expected to create, tracking whether it is realized, and using this understanding to inform future investment. Value management is what allows the enterprise to direct its AI investment on the basis of evidence rather than assumption, learning which investments create value and which do not, and adjusting accordingly. Without value management, the enterprise cannot know whether its AI investment is justified, and it cannot improve its allocation over time. This connects the operating model to the business ownership of value and to the enterprise's broader management of its investments.

Value management must account for the distinctive value profile of AI. The value of AI is not always immediate or easily measured: shared capabilities create value indirectly by enabling many solutions, foundational investments create value over time rather than at once, and some of AI's value is difficult to quantify. Value management must accommodate this profile, recognizing the indirect and long-term value of foundational and shared capabilities rather than judging every investment by immediate, direct returns. An approach to value that demands immediate, measurable returns from every AI investment will starve the foundational capabilities on which the enterprise's AI depends, because their value is real but indirect. Accounting for the distinctive value profile of AI is what allows the enterprise to invest in foundations while still managing value rigorously.

Funding and value management must be sustainable, reflecting the sustainability principle of the operating model. Because Enterprise AI is operated and evolved over time, its funding must be sustained over time, and its value must be managed continuously rather than assessed once. Sustainable funding provides the enduring investment that long-lived capabilities require, and continuous value management ensures that this investment continues to be justified and well-directed as circumstances change. Funding and value management that address only initial delivery, without providing for the sustained investment and ongoing value management that operation and evolution require, undermine the durability that Enterprise AI depends upon.

Funding and value management connect the operating model to the enterprise's strategy and governance. Strategy defines what the enterprise intends to achieve with AI and therefore what its investment should serve; governance defines the risk and accountability within which investment decisions are made. Funding and value management operate within this frame, directing investment toward strategic priorities and making investment decisions accountable and risk-aware. This connection ensures that the enterprise's AI investment is not an isolated financial exercise but an expression of its strategy and a subject of its governance, aligned with what the enterprise intends and answerable to how it governs.

Understood in this way, funding and value management determine whether the enterprise builds durable AI capability and directs its investment toward value. By shifting from project funding to capability funding, sustaining the platform durably, allocating investment according to value and strategy, managing the value that AI creates while accounting for its distinctive profile, and sustaining both funding and value management over time, the operating model ensures that Enterprise AI is resourced to endure and directed toward what matters. These choices are organizational as much as financial, because they determine the character of the enterprise's AI, and making them well is essential to sustaining Enterprise AI as an enterprise capability.

The Enterprise AI Operating Model is the organizational dimension of the Enterprise AI Operating Framework (EAIOF)—the means by which the enterprise arranges its people, structures, and ways of working to deliver and operate Artificial Intelligence as a shared capability. Having examined its structures, roles, decision rights, product orientation, ways of working, talent, and funding, it remains to understand the operating model's role within the framework as a whole. The operating model is the engine of sustainable Enterprise AI: it is what turns the framework's technical structures, governance, and processes into an organization that actually delivers and sustains AI over time.

The operating model gives the rest of the framework an organization to execute it. The architecture and platform define technical structures, but structures require people organized to build and operate them; governance defines policies and accountability, but these require an organization to exercise them; the lifecycle and engineering domains define processes and practices, but these require an organization to carry them out. The operating model provides this organization across all of these, and without it the framework's other elements remain designs without an organization to realize them. This is the operating model's fundamental role: it is the organizational foundation on which the delivery and operation of Enterprise AI depend.

The operating model is the organizational counterpart to governance, with which it is inseparably paired. Governance defines the policies, decision criteria, and accountability that direct the enterprise's AI; the operating model provides the structures, roles, and decision rights through which this direction is exercised. Governance establishes that capabilities must have accountable owners and that decisions must be made within defined rights; the operating model establishes who those owners are and who holds those rights. Neither functions without the other: governance without an operating model is direction with no organization to follow it, and an operating model without governance is organization with no direction. Together they constitute the enterprise's system for directing and organizing its AI.

The operating model provides the organization that the lifecycle and engineering domains require. The Lifecycle Processes define the stages through which AI capabilities pass from conception to retirement, and the Engineering Framework defines the practices through which AI is built; both require an organization—the structures, roles, and ways of working that the operating model defines—to carry them out. The operating model is thus the organizational context within which the framework's processes and practices operate, and the domains that follow build upon it as the organization that executes them. This is why the operating model precedes these domains in the framework: it establishes the organization that the framework's processes and practices assume.

The operating model connects closely to adoption and enablement, which extends its concern with organizational capability to the broader enterprise. The operating model defines the structures and roles that deliver AI and the capability required to staff them; adoption and enablement addresses the broader transformation through which the enterprise's people and culture come to work effectively with AI. The two share the goal of building the human capability that Enterprise AI requires, with the operating model focused on organizing for delivery and adoption focused on enabling the enterprise more broadly. Together they ensure that Enterprise AI is not only technically provided but organizationally and humanly capable.

The operating model is, above all, the foundation of sustainability, which is the quality that distinguishes Enterprise AI from a succession of projects. Sustaining AI over time—keeping it running, evolving it, and continuing to create value from it—is an organizational achievement, dependent on enduring structures, clear ownership, sustained funding, and developed capability. The operating model provides all of these, and it is therefore the primary determinant of whether the enterprise's AI endures as a capability or fades as its initiatives conclude. This is why the operating model is described as the engine of sustainable Enterprise AI: it is the organization that keeps AI running and improving long after it is first built, which is the essence of operating AI as an enterprise capability.

Like the rest of the framework, the operating model must itself evolve. The enterprise's AI adoption matures, its needs change, and the field of Artificial Intelligence advances, and the operating model must adapt accordingly—shifting the balance of centralization and distribution as maturity grows, adjusting roles as needs change, and evolving its ways of working as the collaboration between people and AI deepens. This adaptiveness reflects the continuous-evolution disposition of the framework, and it ensures that the operating model remains suited to the enterprise it serves rather than becoming an arrangement fit only for a stage the enterprise has outgrown. An operating model that cannot evolve becomes, in time, an obstacle to the very AI it was designed to enable.

For these reasons, the Enterprise AI Operating Model should be understood as the organizational engine that makes Enterprise AI sustainable. It gives the framework's technical structures, governance, and processes an organization to execute them; it pairs with governance to direct and organize the enterprise's AI; it provides the organization that the lifecycle and engineering domains require; it connects to adoption in building human capability; and it is the foundation of the sustainability that turns AI from a series of projects into an enterprise capability. By defining how the enterprise organizes for AI, the operating model enables Artificial Intelligence to be delivered, operated, and sustained coherently and at scale—not as something that happens in scattered initiatives, but as something the enterprise is organized to do enduringly.