AI adoption is no longer new for businesses. The next challenge is far more complex: how can organizations move AI beyond tools and pilot projects and make it part of how the business operates end-to-end?

Data from McKinsey illustrates the scale of this gap. In The State of AI in 2025, 88% of respondents said their organizations regularly use AI in at least one business function. However, only about one-third have begun scaling AI, and just 7% report that AI has been fully scaled across their organizations.

Agentic AI shows a similar pattern. While 62% of organizations are at least experimenting with AI agents, only 23% are scaling agentic AI in at least one business function. Across every business function examined by McKinsey, no more than 10% of respondents reported that their organizations had scaled AI agents.

In other words, AI adoption does not automatically translate into AI readiness.

A company may have numerous AI use cases and still lack the processes, data, systems, and governance required for AI to operate in an integrated way at enterprise scale.

This is where the concept of the Autonomous Enterprise becomes relevant.

An Autonomous Enterprise Is More Than a Company That Uses AI

SAP defines an Autonomous Enterprise as an organization capable of continuously sensing what is happening across its operations, reasoning based on business context and predefined rules, and acting across end-to-end processes quickly and at scale without relying on manual coordination at every stage.

However, autonomous does not mean handing every business decision over to AI.

In this model, people define objectives, policies, constraints, and decisions that require judgment. AI agents help execute work, coordinate processes, and take action within predefined boundaries.

SAP summarizes this through three principles:

1. People set the direction, and AI executes.

2. When conditions change, the business can respond as one.

3. Governance enables the business to move faster rather than simply acting as an additional layer of control.

The difference from more conventional AI implementation becomes clearer when we look at the operating model.

This comparison is adapted from SAP's framework distinguishing AI that is simply layered on top of business processes from AI that is embedded into the operational core of the enterprise.

Why Do So Many AI Initiatives Remain Stuck in the Pilot Stage?

The gap between adoption and scale is not simply a question of AI model capabilities.

SAP points to a more structural challenge.

AI is often implemented on top of operating models and system landscapes that were never designed to support coordinated AI execution at enterprise scale. As a result, an individual use case may perform well on its own but become difficult to scale when it has to interact with real-world processes, data, systems, and risks.

SAP identifies three major gaps.

1. Lack of Business and Process Context

Generic AI can read enterprise data, but it does not automatically understand how the business actually operates.

Consider a purchase order.

To an AI model, a purchase order may appear to be a collection of fields and transactional data. Within an actual enterprise, however, a single purchase order may be connected to approvals, compliance checks, goods receipts, accounting entries, supplier payments, and audit requirements.

Each stage may also be governed by policies that determine which actions are permitted next.

Without this business and process context, AI may be able to generate insights or recommendations, but it may not be able to execute decisions safely within the organization's operational constraints.

Readiness questions for businesses:

1. Are core processes documented and standardized?

2. Are decision logic and approval rules clearly defined?

3. Can dependencies between processes be identified?

4. Is it clear which conditions require human intervention?

If the process itself is unclear, giving AI greater autonomy may simply amplify uncertainty.

2. Disconnected Data and Systems

AI can only make decisions based on the context available to it.

The challenge is that enterprise data is often distributed across multiple applications with different definitions, semantic models, and ownership structures. Finance, procurement, supply chain, and other functions may even interpret the same entity or metric differently.

When AI has to reason across these systems, it may effectively be looking at only a partial view of the business.

An output may appear reasonable within one context but create risk or rework when the decision moves downstream into another process.

For this reason, data readiness for an Autonomous Enterprise is not simply about having large volumes of data.

Organizations need to ensure that their data is:

1. Accessible - AI can access the data it actually needs.

2. Consistent - data definitions and structures do not conflict.

3. Contextual - data carries relevant business meaning.

4. Connected - information can be used across systems and processes.

5. Governed - data access and usage are subject to clear controls.

SAP positions shared process models, unified business data semantics, and enterprise-wide AI governance as important components of an operating model in which AI is embedded into the business.

3. Governance Cannot Be an Afterthought

SAP emphasizes that an Autonomous Enterprise requires governance to be embedded directly into execution.

This includes:

1. Policies encoded into execution - agents operate according to the rules, constraints, and objectives defined by the organization.

2. Identity and access control - every action has a clearly defined identity and permission structure.

3. Auditability by design - activities are recorded as part of execution, including who or which agent acted, what authority was used, which data informed the action, and what outcome resulted.

4. Exception handling - when conditions fall outside predefined parameters, decisions are escalated back to people with the context needed to act.

In this context, governance is not an obstacle to autonomy. Governance determines how far autonomy can safely extend.

From AI Assistants to AI Agents

Most enterprise AI usage today still operates within an assistance model.

AI helps people find information, create drafts, analyze data, or summarize reports. These applications remain valuable because they can improve individual productivity.

However, an Autonomous Enterprise requires a shift from AI that helps people perform work to agentic AI that can execute work across processes.

SAP distinguishes the two roles:

Assistants as collaborators

Assistants serve as an interaction layer between people and AI. They help surface insights, understand situational context, coordinate agents toward specific outcomes, and allow people to provide direction or conduct reviews.

Agents as executors

Agents perform specific multi-step tasks using available skills and tools. They can initiate workflows, apply business rules, execute downstream actions, and escalate exceptions when human judgment is required.

This transition matters because processes no longer have to move forward only when someone remembers to initiate the next step.

Execution can become event-driven: conditions change, a signal is detected, a decision is made based on context, and the relevant action is executed.

Sense, Reason, Act: The Operating Loop of an Autonomous Enterprise

SAP uses three capabilities to explain how an Autonomous Enterprise operates.

1. Sense - Continuous Awareness

An organization needs to detect changes as they happen.

Signals may come from transactions, financial positions, inventory, demand, supplier events, workforce data, or external factors.

The key difference from traditional reporting is timing.

Signals do not have to wait for a weekly review or monthly reporting cycle before triggering a response.

2. Reason - Contextual Decisioning

A signal alone is not enough.

AI needs to determine whether the change requires action and which action is most appropriate.

Reasoning requires business context: organizational policies, historical knowledge, objectives, cost structures, customer commitments, compliance obligations, and risk tolerance.

This is what separates a recommendation that merely appears reasonable from a decision that is genuinely appropriate for the enterprise context.

3. Act - Coordinated Execution

Once a signal has been understood and a decision made, agents execute the relevant actions through connected systems.

This may involve executing a transaction, initiating a workflow, adjusting a plan, or escalating an exception to a person.

When sense, reason, and act operate as a continuous loop, an organization can respond to change as one system rather than as a collection of functions operating independently.

Is Your Business Ready?

McKinsey's findings suggest that the challenge facing organizations is no longer simply getting started with AI.

88% of organizations already use AI in at least one function, but only 7% report that AI has been fully scaled. For AI agents, interest also significantly exceeds scale: 62% are at least experimenting, while 23% are scaling agentic AI in at least one function.

This does not mean every organization should pursue full autonomy as quickly as possible.

Quite the opposite. Organizations need to understand which parts of their foundation are not yet ready before expanding autonomy.

This checklist is a practical synthesis of SAP's Autonomous Enterprise framework, not an official SAP maturity model.

The Priority Is Not “More AI,” but a Stronger Foundation

McKinsey found that 80% of respondents said efficiency is an objective of their organizations' AI initiatives. However, organizations generating the most value from AI are more likely to also pursue growth or innovation as objectives. Overall, only 39% of respondents reported EBIT impact from AI at the enterprise level.

These findings demonstrate why an AI strategy should not stop at automation or the number of AI use cases deployed.

An Autonomous Enterprise requires a more fundamental question:

Which processes offer enough business value to transform while also having sufficiently mature process context, data, integration, and governance to support AI execution?

Organizations can assess potential use cases across four dimensions:

From there, organizations can distinguish between use cases that are ready to scale, those that require foundational improvements first, and processes that are not yet appropriate for greater autonomy.

An Autonomous Enterprise Starts with Readiness

The journey toward an Autonomous Enterprise is not a race to deploy as many AI agents as possible.

The more fundamental transformation is building an enterprise that can sense, reason, and act as an integrated system with AI that understands process context, works with connected business data, executes actions across systems, and remains within the governance defined by the organization.

McKinsey's data reinforces the urgency of this transition: AI adoption is already high, but enterprise-scale deployment remains far more limited. The gap suggests that the next challenge is not simply acquiring AI technology, but building the operating foundation that allows AI to generate value consistently at enterprise scale.

As an SAP Partner, Indocyber Global Teknologi helps organizations build and develop the enterprise technology foundation required for SAP-driven business transformation. For organizations evaluating ERP modernization and their readiness to move toward a more intelligent and connected enterprise, you can learn more and discuss your requirements with our team.

[ Consult with our experts -> SAP GROW Fast ]

To stay updated on AI, enterprise technology, SAP, and business transformation, subscribe to the Indocyber email newsletter for our latest insights and updates.