KENNESAW, Ga. | Sep 3, 2026

Why AI begins long before artificial intelligence is talked about

Over the last six months, I have written about what AI leaders should be thinking about.  But I keep returning to a more foundational question: are we focusing too much on AI itself and not enough on the modernization journey that makes AI useful? 

Growth and productivity do not come from models alone. They come from workflows, data, and operating models that have evolved enough for AI to amplify them. In 2026, I found myself navigating three very different systems. On the surface, they have almost nothing in common. Yet every one of these experiences left me thinking about AI. Ironically, none of them taught me anything new about artificial intelligence. They reminded me of something much more fundamental.

That realization became clearer as I reflected on three transportation systems: the New York subway, Indian Railways and Norfolk Southern. They operate at different scales, serve different customers and solve different problems. Yet all three reveal the same truth:

Modernization comes first, AI comes second.

Over the next three newsletters, I'll explore three organizations that solved very different problems—but arrived at the same lesson. AI created value only after decades of modernization had quietly transformed their operations.

Modernization may be the chapter I did not call out clearly enough in my book - AI Leadership Compass: Lead with Clarity.  The AI transformation conversation often misses the real source of momentum behind AI success: the operating environment that existed before AI arrived.

When organizations begin their AI journey, the conversation usually starts with models.

• Should we use ChatGPT?
• Should we build agents?
• Should we buy Copilot?
• Should we fine-tune our own model?

They're important questions. But they often skip an even more important one.

The organizations extracting the greatest value from AI today didn't start with AI. They started years earlier by modernizing how work gets done.

AI needs an ecosystem that can support it before it can perform. Having AI doesn't magically eliminate friction. It learns from the digital infrastructure already in place. It cannot contextualize disconnected information. It performs only as well as that infrastructure allows.  That means organizations need

• AI-ready processes
• Reliable, use-case-relevant data
• Integrated and compatible systems
• Digital workflows designed for intelligent decision making

Without those foundations, AI simply makes inefficient processes happen faster.

Indian Railways

In India, digital reservations made a vast and complex passenger network more accessible. The visible change was the booking experience. Beneath it sat decades of standardization, integration and operational redesign.

Railways are also personal to me. My father’s work in track modernization taught me early that operating intelligence begins with physical signals, like sound, movement, vibration, weather and physical condition—long before those signals become digital data.

My recent experience with Indian Railways reminded me what modernization can achieve in scale. The scale is mind-boggling: Indian Railways supported approximately 7.4 billion passenger journeys during 2025.

Norfolk Southern

In freight rail, modernization is largely invisible. Sensors, inspection systems and AI models help identify risk while trains remain in motion. They are quietly preventing problems before they occur. The contribution of intelligence powering AI is rarely visible to the public. They communicate early warning signals about emerging problems so teams and operating systems can react early to mitigate risk. 

Each system demonstrates a different outcome of modernization:

• In New York, customer friction disappears.
• In Indian Railways, complexity is orchestrated at scale.
• In freight rail, risk must become visible early enough for intervention.

Different systems. Different outcomes. The same underlying principle: intelligence creates value only when the operating environment is ready to use it.

As I reflected on these experiences, a pattern emerged. Talking about how to operationalize AI is one thing; seeing modernization unfold across transportation systems shows how leaders create the conditions for AI to operate at scale.

None of these organizations began with AI. They evolved as new technologies, customer expectations, operational pressures, and delivery pain points created a need to modernize. Over time, they built on the strengths of earlier platforms and focused on one priority-

As I reflected on these experiences, a pattern emerged. Talking about how to operationalize AI is one thing; seeing modernization unfold across transportation systems shows how leaders create the conditions for AI to operate at scale.

None of these organizations began with AI. They evolved as new technologies, customer expectations, operational pressures, and delivery pain points created a need to modernize. Over time, they built on the strengths of earlier platforms and focused on one priority- They modernized:

• Customer experiences
• Payment systems
• Operational workflows
• Infrastructure
• Decision processes

AI transformation is not an event. It is the visible outcome of years of modernization. That view is echoed in a World Economic Forum white paper:

From my viewpoint, I see modernization as a progression, not a technology project. It is less about adopting tools and more about building the conditions that allow intelligence to create value.

Modernization Ladder

Digitalization → Standardization → Integration → Context → Intelligence → Personalization → Autonomy

Previously, I have written about AI adoption, AI fluency, and context engineering. Context does not appear by accident; it emerges from modernized systems. Personalization does not happen because of better prompts. It happens because organizations have created an environment where AI has access to the right information, at the right time, within the right business process.

Prompts help search, analyze, and execute the commands we present; they do not replace the foundations that make reliable, original work possible.

The question we keep asking may be too narrow.

Instead of asking only, “How do we scale AI?” leaders should ask, “What must the organization become capable of doing before AI can scale?”

Organizations leading with AI are not necessarily leading because they adopted better models. Many are leading because they modernized earlier.

This article introduced three experiences that changed how I think about modernization and AI readiness.

Over the next couple of months, I will unpack each story individually and explore what it reveals about:

• How modernization removes customer friction
• How operational redesign enables scale
• How invisible intelligence turns risk signals into timely action

Final Learning point: AI does not create organizational intelligence. It amplifies the organization’s existing capacity to turn signals into context, decisions and action.

_{Reference: AI Leadership Compass, Chapter 16, “AI Value and Impact,” and Chapter 12, “Tech-as-a-Service: The AI Backbone.” AI Leadership Compass: Lead with Clarity: 7 Moves That Power AI Transformation: 9781736706985: Sarathy PhD, Priya: Books}