How Leaders Should Sequence Systems Modernization and Automation

The executive decision is not whether the organization should pursue ERP or AI as competing technologies. It is how leadership should sequence process redesign, systems modernization, data governance, and automation.

Opening Thesis

Artificial intelligence will not repair unreliable transactions, fragmented systems, undefined processes, or unclear ownership.

Neither will ERP by itself.

The executive decision is not whether the organization should pursue ERP or AI as competing technologies. It is how leadership should sequence process redesign, systems modernization, data governance, and automation so that each investment strengthens the operating model rather than institutionalizing its weaknesses.

For some organizations, controlled AI experimentation can proceed now.

For others, the first responsible investment is repairing the system of record and the processes around it.

THE GOVERNING QUESTION The governing question is not simply: Where can the company use AI? It is: Is the operating foundation reliable enough for consequential automation to be trusted?

The Business Situation

Organizations are under pressure to adopt artificial intelligence.

Boards want to understand the opportunity. Employees are experimenting with new tools. Technology providers are embedding AI into enterprise applications and presenting automation as the next source of productivity.

Adoption is broad, but enterprise integration remains limited.

McKinsey’s 2025 global survey found that 88 percent of respondents’ organizations regularly used AI in at least one business function, while only approximately one-third had begun scaling AI across the enterprise.^[1]

The distinction matters.

Using an AI tool is not the same as integrating AI into a reliable operating model.

For most organizations, AI remains primarily an automation and decision-support capability – not a substitute for executive judgment. It may prepare an analysis, recommend an action, classify information, route work, or automate part of a workflow.

Leadership remains responsible for setting objectives, evaluating tradeoffs, allocating capital, accepting risk, and deciding what should happen next.

The better executive question is therefore not where AI could be used.

It is which processes are reliable enough to automate – and which operating foundations should be repaired first.

Executive Perspective

AI scales the operating system to which it is connected.

If that system contains reliable transactions, standardized processes, clear ownership, disciplined controls, and governed data, automation can improve speed, consistency, and visibility.

If the organization depends on disconnected applications, uncontrolled spreadsheets, inconsistent master data, undocumented workarounds, outdated equipment, or employees manually reconciling conflicting reports, automation can scale those weaknesses as well.

In a manual process, an experienced employee may recognize that a transaction does not make sense. The employee may investigate an inconsistency, challenge an assumption, or correct information before it moves further through the organization.

An automated error can move through purchasing, inventory, billing, forecasting, customer service, and financial reporting before anyone recognizes that the underlying logic was wrong.

The organization receives an answer faster while becoming less certain that the answer is reliable.

The objective of systems modernization is not merely to prepare for AI.

It is to create an operating environment in which people and technology can perform reliably.

ERP Is Not the Cure

The phrase “ERP before AI” should not be interpreted as a universal requirement to replace the company’s enterprise platform.

ERP is not a cure for unclear ownership, poorly designed processes, weak controls, or inconsistent management decisions.

A new platform can institutionalize confusion as effectively as AI can accelerate it.

Before selecting or redesigning technology, leadership must understand:

• What process the organization is trying to improve.
• Who owns the process and its outcomes.
• Which transactions and information are required.
• What controls and approvals are necessary.
• How exceptions should be identified and resolved.
• How performance and reliability will be measured.
• Whether the expected business value justifies the investment.

Some organizations need a new ERP.

Others need better configuration, integration, master-data governance, process discipline, training, equipment, or accountability.

The investment should follow the operating diagnosis – not precede it.

The Foundation People and AI Both Require

Reliable Information

Employees cannot make sound decisions when customer, product, inventory, pricing, supplier, or financial information is incomplete or contradictory.

Neither can AI.

Organizations need clear definitions, reliable systems of record, disciplined master data, and accountable ownership.

Without them, employees spend time locating information, reconciling reports, correcting transactions, and debating which version is accurate.

Automation may make the confusion move faster. It does not resolve the disagreement.

Standard Processes and Clear Accountability

Leadership should not automate a consequential process it has not adequately defined.

When departments perform the same activity differently, employees rely on memory, informal relationships, and individual judgment. Those workarounds make performance difficult to measure, improve, transfer, or scale.

Before automation, leadership should understand:

• What initiates the process.
• Who owns each decision.
• What information is required.
• Which approvals matter.
• How exceptions are managed.
• How completion is verified.

A technology team may configure an application.

It should not determine purchasing authority, pricing logic, accounting policy, inventory ownership, customer terms, or financial-reporting definitions.

Those decisions belong to operating and executive leadership.

Effective Infrastructure and Controls

Operating foundations are not purely digital.

Modern software cannot compensate for unreliable networks, outdated equipment, disconnected production systems, or physical workflows that prevent accurate and timely data collection.

Employees already absorb the cost through downtime, duplicate entry, weak interfaces, manual workarounds, and information that does not move with the work.

Automation also increases the importance of control.

As software recommends or initiates activity, leadership must understand:

• What the system is permitted to do.
• What requires human review.
• How exceptions are identified.
• How activity is monitored.
• How incorrect actions can be interrupted or reversed.
• Who remains accountable for the outcome.

NIST’s Artificial Intelligence Risk Management Framework emphasizes incorporating trustworthiness into the design, development, use, and evaluation of AI systems.^[2]

Governance should be part of the operating design – not an addition after deployment.

The Modernization Decision Standard

Leadership should evaluate each proposed AI or automation use case against the condition of the underlying operating process.

Operating Condition	Leadership Decision
Reliable process, reliable data, and limited consequences	Proceed with controlled experimentation. Define the intended outcome, accountable owner, permitted use, review requirements, and success measures.
Reliable process but fragmented applications or interfaces	Improve integration and information flow before scaling automation. The process may be sound even when the architecture creates unnecessary friction.
Inconsistent process or unclear ownership	Redesign the process and assign accountability before making a material technology investment. Automation should not become a substitute for executive decisions leadership has avoided.
Unreliable transactions or master data	Stabilize the system of record before relying on AI for consequential analysis or action. Faster processing does not improve defective inputs.
ERP materially limits reporting, controls, scalability, or execution	Evaluate modernization or replacement based on business value, implementation risk, operating capacity, and total cost – not technology preference.
High-consequence use case without adequate controls or traceability	Do not automate consequential decisions until leadership can explain, monitor, challenge, and govern the resulting activity.

A Practical Modernization Sequence

Organizations do not need to delay every AI experiment until every system is perfect.

They do need to sequence modernization responsibly.

1	ASSESS	Understand how work actually moves through the organization. Identify fragmented systems, physical constraints, manual corrections, inconsistent data, unclear ownership, recurring failure points, and decisions that depend excessively on individual knowledge.
2	STABILIZE	Restore transaction discipline, reporting reliability, process ownership, infrastructure availability, control effectiveness, and management confidence. The organization should first be able to trust the basic activity upon which future automation will depend.
3	TRANSFORM	Redesign the process before automating it. Modernize ERP, equipment, integrations, workflows, and data structures where current conditions materially limit performance, visibility, control, or scalability.
4	OPTIMIZE	Apply AI and other automation to processes that are reliable, repeatable, controlled, and economically meaningful. Technology should improve a defined business outcome – not merely demonstrate technical capability.
5	SCALE	Expand only after the organization can monitor performance, cost, security, controls, employee impact, operating risk, and realized value. Scaling should be earned through evidence.

SEQUENCE BEFORE SCALE This sequence does not delay innovation. It prevents the organization from institutionalizing weak processes under the appearance of innovation.

Practical Implications

AI may provide the urgency required to fund systems improvements that leadership has deferred for years.

That may become one of its most important near-term contributions.

A company may justify an ERP upgrade, infrastructure investment, process redesign, equipment modernization, integration project, or data-governance initiative because leadership wants to become AI-ready.

The first return, however, may appear before consequential AI is deployed.

• Employees may close the books faster.
• Operating teams may locate information more easily.
• The organization may reduce duplicate work.
• Inventory accuracy may improve.
• Customer issues may be resolved sooner.
• Controls may become stronger.
• Leadership may make decisions with greater confidence.

McKinsey’s research also suggests that the organizations reporting the greatest AI impact are distinguished by practices such as workflow redesign, senior-leadership ownership, defined human-validation processes, data and technology infrastructure, and embedding AI into business processes.^[1]

The technology matters.

The operating model determines whether it creates value.

AI may be the catalyst. The operating improvement is the value.

Executive Takeaway

EXECUTIVE TAKEAWAY ERP before AI is not a rigid technology rule. It is a decision standard. Organizations should not automate consequential processes until leadership understands and trusts the transactions, data, infrastructure, controls, ownership, and operating logic beneath them. A new ERP cannot repair weak leadership or undefined processes. AI cannot make unreliable information trustworthy. Leadership must first create an operating foundation that employees can use, managers can govern, and executives can rely upon. Companies may modernize because they want to adopt AI. They should recognize that their employees needed the same operating foundation long before AI arrived.

About CFO’s Office Executive Perspectives

Executive Perspectives is the thought-leadership publication of CFO’s Office.

Each perspective examines an enduring executive leadership principle through the lens of current business conditions, helping CEOs, boards, investors, family offices, private equity firms, business owners, and executive leaders improve decision quality.

Rather than focusing on technical finance alone, Executive Perspectives examines how financial leadership strengthens enterprise value through Financial Clarity, Executive Confidence, and Disciplined Execution.

About the Author

Kyle Casey, CPA, is the Founder of CFO’s Office and an operating executive, former public-company Chief Financial Officer, and former Interim Chief Executive Officer. He advises CEOs, boards, business owners, family offices, and private equity stakeholders on executive financial leadership, enterprise transformation, governance, liquidity, finance modernization, and operational discipline.

His career spans public companies, consumer products, manufacturing, distribution, gaming, hospitality, entertainment, government, and regulated operating environments.

His leadership philosophy is grounded in a simple principle: Finance exists to improve decision quality. By building reliable operating foundations, disciplined processes, and decision-quality information, he helps organizations transform vision into sustainable results.

Sources and Notes

[1] McKinsey & Company, “The State of AI in 2025: Agents, Innovation, and Transformation,” November 5, 2025. The survey reports that 88 percent of respondents’ organizations regularly use AI in at least one business function, while approximately one-third have begun scaling AI across the enterprise. It also identifies workflow redesign, senior-leadership ownership, human-validation processes, data infrastructure, and technology infrastructure among the practices associated with stronger AI outcomes. Source

[2] National Institute of Standards and Technology, “Artificial Intelligence Risk Management Framework 1.0,” January 2023. NIST describes the voluntary framework as a means of incorporating trustworthiness considerations into the design, development, use, and evaluation of AI products, services, and systems. NIST currently notes that AI RMF 1.0 is being revised. Source

A QUESTION WORTH ADVANCING If this perspective prompted a question about your organization, board, or portfolio company, I would welcome the conversation.

Kyle Casey, CPA
Founder | Chief Financial Officer
CFO’s Office
cfosoffice.com