In brief
Across boardrooms globally, CEOs are operating under some of the most intense performance pressure seen in decades. Growth expectations remain elevated despite geopolitical instability and economic uncertainty. Investors are demanding visible financial discipline and operational resilience, while simultaneously expecting organizations to accelerate innovation and improve market agility to boost productivity and returns. Against this backdrop, one mandate from boards has become increasingly explicit: leverage agentic AI to deliver measurable value.
Yet despite years of investment, most organizations remain trapped in pilot mode. Proof of concept continues to proliferate and AI activity across the board accelerates rapidly, but enterprise impact and value creation are not moving at the same pace.Â
EY’s 2025 Work Reimagined Survey captures this disconnect clearly: while 88% of employees now use AI at work, only 28% of organizations are positioned to translate that activity into meaningful business outcomes. In this article we explore how a missing link could be strategic design and execution of AI, including agentics: governed AI agents that can coordinate workflows, decisions and actions across functions.
This article – the first in the thought leadership series on the Agentic Enterprise – examines why AI has so far failed to scale at the enterprise level, what is missing in today’s operating model and technology architecture, and what critical capabilities must be in place for organizations to compound AI’s full value potential.
The thought leadership series C-suite strategies for AI-enabled enterprise reimagination will cover the following topics:Â
Chapter 1
Most organizations are doing AI, but very few are transforming with AI.
Every function – finance, supply chain, HR, manufacturing or customer service – has its own growing list of automation or insights initiatives. These localized projects improve functional efficiency but rarely scale to enterprise-wide impact. Organizations are investing significantly in more models and tools, but the sources of delay, rework and decision friction that shape enterprise performance remain largely untouched.
In practice, we see most organizations deploying AI in ways that mirror existing organizational boundaries rather than focusing on the flow of value across them. Functional leaders naturally tend to anchor AI ambition around near‑term, demonstrable returns within their sphere of influence, resulting in activity to optimize existing processes and ways of working. AI is applied primarily for cost‑out and automating predictable tasks to improve productivity rather than reimagining for end-to-end organizational agility and overall cost‑to‑serve. While the in-function initiatives deliver some productivity gains and help stabilize core functional operations, they are structurally limited in enterprise impact.
This is because only a minority of the overall value creation potential sits within individual functions and the majority, often as much as 70–75%, is trapped in the silos between them.
How EY-Parthenon can Help
Discover EY-Parthenon’s AI-powered deal technology platforms that drive real-world results across strategy, transactions and transformation.
Read moreUnlocking this trapped value requires activating cross-functional workflows at scale (we call these enterprise value streams) that will address decision-making friction, approval latency, coordination overhead, analysis duplication, rework loops and fragmented accountability.
Â
These friction factors rarely appear clearly on organizational charts or financial statements, yet collectively they shape enterprise responsiveness, execution speed, operating cost, customer experience, and ultimately value creation.
Â
This is the hidden “75% gap” and remains largely untouched by today's AI efforts. This mismatch explains why organizations are doing more AI every year but not seeing the compounding effect of AI in enterprise results.
Â
A recent automotive original equipment manufacturer (OEM) transformation value estimation led by EY-Parthenon illustrates the scale of the opportunity. Total AI- and digital-enabled value creation potential exceeded US$2.4b, of which nearly US$2b came from enterprise value streams.
Â
Why enterprise AI fails to compound value
Â
The root cause lies in the architecture of the modern enterprise itself. Over decades, organizations and the systems supporting them evolved vertically around functions, reporting hierarchies and control structures. enterprise resource planning (ERP), customer relationship management (CRM), human resources information systems (HRIS), product lifecycle management (PLM) and workflow platforms were designed primarily to optimize functional execution, governance and control. But enterprise value does not flow vertically. It flows horizontally across customer journeys and product lifecycles through cross-functional operational value streams such as concept-to-launch, order-to-cash, procure-to-pay and service incident-to-resolution.
Â
This creates a fundamental structural tension: people, systems, key performance indicators (KPIs) and governance are organized vertically, while value creation flows horizontally. Where these intersect, friction emerges. In most enterprises, a substantial proportion of effort is consumed not for value creation itself, but for managing the friction between silos.
Â
In the recent automotive OEM example, the Incident-to-Resolution process spans across eight functions, touches a number of enterprise systems and requires significant human intervention, with 15-25% of the effort in warranty management spent on internal coordination activity rather than solving customer or operational problems directly.
Traditional AI initiatives rarely address this problem because they inherit the same structural boundaries as the organization itself. This is where agentic AI – governed AI agents capable of stitching together the cross-functional workflows across systems, data and functional silos, with humans in the loop – becomes central to value creation.
Five barriers preventing AI from scaling enterprise value
Most AI programs focus on improving existing functional activities rather than redesigning how value moves across the organization. This creates incremental efficiency improvements but limited enterprise agility. As a result, organizations optimize locally while enterprise-level friction persists globally.
2. AI is being wrapped around existing processes
Many organizations are layering AI onto current operating models rather than redesigning them. This matters because AI wrapped around fragmented workflows simply accelerates fragmentation. Instead of fundamentally improving enterprise flow, AI risks becoming a sophisticated incremental efficiency layer sitting on top of structurally inefficient processes. The result is faster execution of the same broken coordination model.
3. Enterprise data and systems remain fragmented
Most enterprise architectures were never designed for AI-native cross-functional coordination. Critical information remains fragmented across ERP, CRM, engineering, operational and supply chain platforms. Teams still spend enormous effort reconciling data, validating assumptions and resolving inconsistencies across systems that were designed independently around functional priorities. This fragmentation creates hidden operational drag that scales with organizational complexity.
4. Organizations lack a shared intelligence and context layer
This is the most consequential barrier – and arguably the least understood. To operate effectively across systems, functions and workflows, AI agents require organizational context. Without this shared context layer, AI agents can analyze data in silo but cannot coordinate decisions coherently across enterprise value streams. This intelligence layer – enabled through ontologies, semantic models, knowledge graphs and contextual reasoning – is not optional infrastructure. It is the architectural foundation that allows AI to compound value across the enterprise. Without it, AI scales activity. With it, AI scales decision quality and enterprise value flow.
5. Governance and operating models have not evolved
Despite rapidly increasing AI investment, most organizations still govern AI through fragmented functional structures. In most organizations, AI activity is measured while enterprise impact is not; governance remains episodic rather than continuous; and accountability remains vertical rather than cross-functional. As a result, AI programs scale experimentation faster than they scale enterprise performance.
Chapter 2
The Agentic Enterprise uses AI-driven orchestration to coordinate workflows, decisions and value streams across the organization with humans guiding oversight.
The next phase of enterprise transformation will not be defined by isolated AI tools. It will be defined by AI-enabled orchestration. This is the emergence of the Agentic Enterprise. In the Agentic Enterprise, AI no longer operates only as an assistant to individual workers. AI agents coordinate workflows, decisions and operational activity across functions, systems and value streams. AI begins to act horizontally rather than vertically. Instead of relying on manual coordination between siloed teams, organizations increasingly use AI agents to orchestrate cross-functional workflows, coordinate decisions across systems and surface risks and bottlenecks in real time.
The transformation lies in enabling value to flow more intelligently across them. Intelligent orchestration, with humans increasingly acting as supervisors, conductors and exception managers rather than process coordinators, is more realistic than wholesale structural redesign.
The next step is clear: mapping the foundational operating model architecture that must be built to make that progression possible. This path runs through a structured intelligence layer that gives AI agents the context, consistency and autonomy to act at enterprise scale. Building that architecture is the strategic priority that determines which organizations get there first.
The seven-layer operating model of the Agentic Enterprise
The Agentic Enterprise requires a new enterprise operating architecture. In this model, AI becomes an embedded strategic enterprise capability woven across seven interconnected operating model layers.
1. Infrastructure layer
At the foundation sits scalable cloud, compute, cybersecurity and data infrastructure capable of supporting enterprise-wide AI orchestration securely and reliably.
2. Enterprise Systems of Record layer
ERP, CRM, HRIS, PLM and operational systems continue to manage transactional integrity, compliance and core enterprise records. However, their role increasingly shifts from being systems of workflow logic toward systems of record.
3. Intelligence and context layer
This is the most strategically important layer – and the one most organizations have not yet built. Here, shared enterprise meaning is created through semantic models, ontologies, knowledge graphs, and contextual relationships. This layer establishes common understanding across customers, products, suppliers, financial entities, operational states and risks. Without this layer, agents may act, but they cannot act coherently across functions. The intelligence layer is the precondition for enterprise-wide orchestration.
4. Agent and orchestration layer
Above the intelligence layer sits the orchestration layer, where AI agents coordinate workflows, decisions, and actions dynamically across systems and functions, orchestrating value flow across enterprise value streams.
5. Governance, trust and control layer
As AI becomes embedded into enterprise operations, governance becomes continuous rather than episodic. Guardrails become embedded into the architecture itself rather than added after deployment.
6. Workflow and decision layer
This layer governs how decisions are executed, escalated and monitored operationally across the organization. Humans remain embedded where judgment, accountability, ethics or regulatory oversight are required.
7. Business outcome layer
At the top sits the enterprise outcome layer – where AI continuously optimizes strategic objectives such as growth, cash flow and operational resilience. This is ultimately where AI becomes measurable as enterprise performance.
Strategic implications for CIOs and CFOs on enterprise technology investments
As agentic AI and intelligence layers mature, traditional enterprise systems increasingly evolve toward stable systems of record rather than heavily customized workflow engines. Critical orchestration, business logic and cross-functional coordination progressively move into the intelligence and agentic layers. For CIOs and CFOs, this changes the technology investment thesis fundamentally, decreasing customization and increasing flexibility. This could materially reduce the brittleness, complexity and cost of future enterprise transformation programs.
Implications for CEOs and boards on the enterprise operating model
The implications extend far beyond just technology architecture: as enterprises evolve toward agentic operating models, leadership itself changes.
Today, executive teams primarily manage through vertical functions and retrospective reporting structures. In the Agentic Enterprise, leadership increasingly steers the organization through real-time enterprise signals tied directly to strategic outcomes. Decision-making becomes faster because AI continuously breaks down issues, surfaces dependencies and orchestrates workflows across functions.
New leadership constructs emerge: Value Stream Leaders, Agent Owners and Enterprise Orchestrators. Governance shifts from silo oversight toward continuous enterprise steering. Functional expertise, regulatory control and domain depth remain essential. The transformation lies in how work, information and decisions flow across them.
How EY-Parthenon can Help
AI Strategy Services team can enable better and faster ROI for new products and services and help accelerate M&A. Find out more.
Read moreThe path to the Agentic Enterprise does not begin with a transformation program. It begins with a single, well-chosen outcome. If AI has been deployed across your functions for several years but enterprise performance has not materially shifted, the critical question is whether AI is being applied to the right unit of value. The Agentic Enterprise reference reframes AI as a mechanism for orchestrating value flow through the enterprise – which is where meaningful returns emerge.
Â
For most organizations, the practical starting point is narrower than the ambition. We recommend four steps to begin the transition:
The next frontier of enterprise value
The organizations that win with AI will not be the ones that deploy the most pilots, build the most models, or automate the most tasks. They will be the ones that use AI to redesign how the enterprise itself operates. The real prize is not the 25% of value trapped inside functions – it is the 75% lost between them.
In the months and years ahead, the question will no longer be whether organizations adopted AI, but whether AI fundamentally changed how value flows through the business. This is the crucial shift that CEOs now face: from using AI to optimize work to using AI to orchestrate the enterprise. The gap between those two ambitions will define the next generation of market leaders.
Are you ready to close the 75% gap?
About the authors
The following individuals contributed to this white paper: Will Auchincloss, EY-Parthenon; Nils Melcher, EY-Parthenon; Suraj Ramprasad, EY-Parthenon and Gaurav Batra, EY.
Frequently asked questions about agentic AI and enterprise value
AI adoption is accelerating, yet enterprise value creation remains limited. Close the 75% value gap by using agentic AI to connect workflows and unlock cross-functional enterprise value at scale.
Related content
Geostrategic Analysis: June 2026
Read the June 2026 Geostrategic Analysis for our take on geopolitical developments and the impact of these political risks on international business.
Top 10 geopolitical developments in 2026
Geopolitical volatility and uncertainty will persist in 2026. There are three geostrategic actions to build resilience and set strategy. Learn more.
How the productivity reset will redefine value
Discover how AI and megatrends will transform productivity, drive new value measures, smarter strategies, and human-machine collaboration.
