Top Down Is the Next Phase of Enterprise AI Strategy

AI adoption began quietly at most organizations. In many cases, it started with a single employee who shared with their co-workers an AI tool that they were secretly using to save time. Or perhaps it was a developer who created an AI chatbot to answer training questions at a team meeting. It might even have been a manager who used their department budget to acquire for employees the same AI tool that they already used at home.

Then, seemingly overnight, spiderwebs of AI tools spread throughout entire organizations. Often, the tools increased productivity and decreased errors. But they also created new problems.

As a result, enterprise AI has now entered a new phase. After several years of pilots, proofs of concept and department-level experimentation, organizations are under pressure to turn AI activity into measurable, repeatable business impact. In fact, only 1 in 10 companies say they have fully embedded AI capabilities that consistently deliver value, McKinsey & Company said in its January 2025 report, “Superagency in the Workplace: Empowering People to Unlock AI’s Full Potential.”

In addition to tools, organizations need the workforce and structure to make AI meaningful, according to enterprise software company ServiceNow. In a November 2025 blog post, it explained that while frontline employees often are driving AI adoption, organizations aren’t structurally ready at the top.

“Agentic AI, a new generation of intelligent systems that can reason, plan and take action on their own, is creating a new class of employee-led innovation,” author Ashmita Shrivastava wrote. “The same momentum also challenges organizations to strengthen their AI governance, AI strategy and culture, turning early experimentation into lasting advantage.”

The Problem with Fragmentation

The early wave of AI adoption was largely driven by experimentation and local optimization, according to David Marco, president of EWSolutions, a strategic advisory and implementation firm focused on data and AI governance.

“That worked when the stakes were low. At scale, it breaks down,” Marco told The Forecast in an interview. “AI systems begin to influence material business decisions, and the organization can no longer tolerate ambiguity around data ownership, decision rights or risk accountability.”

Organizations that continue with a fragmented approach risk the spread of “shadow AI,” suggests an April 2026 survey of 1,000 IT decision-makers by Freshworks.

Shadow AI happens when employees use unauthorized AI tools in their daily work, usually to automate tasks. While 92% of IT leaders say they have full visibility into AI tools across their environment, 71% admit that unapproved AI use is already common, according to Freshworks, which said 86% of organizations have experienced “negative incidents” as a result of unapproved AI.

One significant consequence of shadow AI is tool sprawl, where multiple tools are created serving the same purpose. Fragmentation also means that organizations lack a centralized list of tools that are being used. That can lead to other risks, such as:

• Security exposure: Each new tool used within an organization creates an opportunity for cybercriminals to access the organization’s data and infrastructure. When IT does not know about a tool, they cannot actively protect the endpoint, which amplifies the risk.

• Compliance gaps: In regulated industries such as finance and healthcare, organizations must ensure that all data and tools comply with specific mandates. When the compliance department is unaware of all AI uses, it can’t effectively enforce protocols, which opens up the organization to fines, data loss and reputation damage.

• Redundant spending: When the organization lacks a formal process and a complete inventory of tools, departments often waste time and money purchasing or developing tools that already exist within the organization.

IT Leaders’ Evolving Role in Enterprise AI Strategy

CIOs, CTOs and other senior leaders are leading the shift from fragmented innovation to enterprise-wide outcomes, according to Marco, who said that top-down AI governance is defining the next phase of enterprise AI strategy. This is the path forward, he explained, because it forces clarity in organizations that have otherwise been operating on implicit assumptions.

“What many organizations miss is that scaling AI is not a technology challenge. It is a leadership and operating model challenge,” Marco told The Forecast. “Budgets, data strategies and governance structures only work when they are aligned to a clear decision framework that holds under pressure.”

In a February 2026 editorial for CIO magazine, Sean Heuer, chief customer officer at Resolve Systems, explained that the next generation of AI leaders is less focused on vision than operations. Instead of usage, he observed, they’re measuring success by how reliably AI delivers value within the enterprise.

Key responsibilities include embedding AI into production workflows; establishing guardrails for responsible use; defining metrics for impact and ROI; and coordinating how models, data and platforms are used across the organization.

“This iteration of the role looks less like a standalone innovator and more like a cross-functional orchestrator,” Heuer said. “You can see the emphasis shifting from asking what AI can do to ensuring that AI works safely, as well as in harmony with business goals.”

Four Steps Toward Top-Down AI Management

Niranjan Krishnan, head of AI solutions at FPT Software — a global technology and IT services provider headquartered in Vietnam — told The Forecast that organizations must move beyond “the AI valley of death,” which is what he calls the fragmented proliferation of AI tools that exists inside so many organizations.

By taking the following four steps, Krishnan helped FPT Software achieve business value from AI at scale:

1. Establish production-ready platform infrastructure. The first step is moving away from isolated experiments and toward a shared environment that makes deployment a standard protocol. Currently, a vast majority of corporate generative AI projects fail to deliver measurable P&L impact. By using top-down governance, organizations can begin to bridge this infrastructure gap, Krishnan said.

2. Create a centralized portfolio discipline: Next, apply a rigorous measurement framework that ties micro-level wins to macro-level financial outcomes, Krishnan advised. Without this framework, task-level success often masks system-level failures. In one study, the National Bureau of Economic Research found that this approach yields productivity gains of 14-34% for individual users.

3. Embrace strategic workforce realignment: With entry-level job postings in certain sectors declining as AI absorbs junior tasks, leaders must proactively restructure roles to focus on judgment-based work rather than manual execution, Krishnan said.

4. Promote unified data and governance: Organizations must also overcome departmental silos to build a unified data strategy, according to Krishnan, who said centralized accountability is required to ensure AI-generated “workslop” does not create compounding rework for downstream teams.

How to Balance Executive Direction with Innovation at the Edges

As organizations move to the next phase of AI, it’s vital to define the goal that you’re moving toward, stressed Marco. He explained that centralization is not the goal; rather, organizations must move toward accountability for decision-making.

“Top-down strategies succeed when they explicitly define who owns the decisions AI is informing, what level of risk is acceptable and how those decisions hold under scrutiny,” Marco told The Forecast. “Without that, organizations end up with technically functional models producing outputs that cannot be trusted or defended.”

Related:

The CIO’s Guide to Unlocking Scale with Enterprise-Grade GenAI;

Inference Is Coming Home: The Quiet Reversal from Cloud-Only to On-Prem + Edge AI

Beyond the Hype: Scaling GenAI into a Competitive Edge

Futurist Mike Bechtel Sees AI Shifting From Push to Pull in 2026

Agentic AI Reconfigures Customer Service  

AI Trends to Watch in 2026: Finding the Right Compute Platform for Each Workload  

Study Shows Big Uptake of Enterprise AI and Cloud Native Technologies