The Value Creation Imperative With AI

Across boardrooms, one topic dominates executive conversations these days: generative AI (GenAI). Over the past three years, companies have started to embrace this technology with remarkable speed. Pilots have multiplied, copilots have been rolled out, and proofs of concept have captured the imagination of leaders. Yet the (measurable) results have often been underwhelming.

This is the GenAI paradox: widespread enthusiasm but limited impact. According to McKinsey research, nearly 8 in 10 companies globally have adopted GenAI in at least one business function, but roughly the same share reports no material effect on their bottom line. The challenge may not stem from ambition or investment, but rather from the approach taken in implementing AI, often added to existing processes instead of being used to fully redesign core operations and ensure value creation.

So far, most deployments have been horizontal: enterprise-wide copilots and chatbots that boost individual productivity but deliver diffuse and hard-to-measure gains. Tools like these help employees save time, summarize documents, or draft communications—real improvements, but typically incremental and difficult to translate into bottom-line impact.

By contrast, vertical applications, those tailored to specific functions such as logistics, or procurement, have struggled to scale. They require custom development, integration across systems, and a deep understanding of business logic. As a result, fewer than one in ten such use cases ever move beyond the pilot phase.

In many organizations, this imbalance has left AI stranded on the margins of true transformation. Fragmented initiatives, data silos, and cultural apprehension have prevented companies from realizing the full potential of GenAI. 

If applied well and used to redesign business processes, much of the value can come from AI agents.

The Rise of Agentic AI

AI agents, if not new, are offering a way to bridge the gap between potential and performance. Unlike traditional generative systems, AI agents can reason, plan, and, more importantly, can act toward defined goals. They combine large language models with memory, orchestration, and integration layers that allow them to operate autonomously across workflows and systems.

We are already seeing outstanding examples. In banking, AI agents are already helping teams modernize legacy technology stacks, cutting development time by more than half. Agents are drafting credit-risk memos autonomously, reducing turnaround times by up to 60 percent. In research and analytics, they detect data anomalies and synthesize insights faster than teams of analysts could.

According to our "State of AI 2025" report, AI agents are increasingly being adopted across industries worldwide, more often in technology, media and telecommunications, and healthcare. Locally, Microsoft reported that 41% of Mexican business leaders are already utilizing AI agents to automate workflows and business processes, and 89% of companies plan to expand their adoption. 

The impact goes far beyond efficiency. Agents enable operational agility—they can adapt to changing data in real time, reconfigure processes, or detect risks before they escalate. They also unlock new sources of growth, powering adaptive customer experiences, personalized product recommendations. In industrial contexts, agents embedded in connected products or equipment could monitor usage, detect performance thresholds, and autonomously unlock features or trigger maintenance actions.

In short, agentic AI transforms AI from a reactive assistant into a proactive collaborator—one capable of redefining how value is created.

From Automation to Reinvention

To capture this opportunity, organizations have to move beyond “plug-and-play” pilots and undertake a deeper transformation: reinventing business processes around agents. When agents are merely inserted into legacy workflows, they serve as faster assistants. But when processes are redesigned to harness their strengths (parallel execution, adaptability, personalization, and elasticity) the results are transformative.

Consider a customer service operation. In a conventional setup, agents (human or digital) respond reactively to inquiries. With agentic AI, those same systems can proactively detect common customer issues by monitoring patterns across channels, anticipate likely needs, initiate resolution steps automatically (such as issuing refunds, reordering items, or updating account details) and communicate directly with customers via chat or email Human employees are elevated from task execution to exception handling and quality assurance.

A New Foundation for Scale

Scaling agentic AI, however, demands more than technology. It requires a new architectural paradigm and a new operating model.

On the technical side, the emerging concept of the agentic mesh offers a blueprint: a modular, governed architecture that allows multiple agents — custom-built or off-the-shelf — to collaborate securely across enterprise systems. It enables composability, vendor neutrality, and continuous evolution as models improve.

But the bigger challenge is – as so often – change management. As agents evolve from passive copilots to proactive decision-makers, organizations must rethink how humans and machines coexist. How do we define accountability when a system acts autonomously? How do we earn trust from employees and customers? 

These are not purely technical questions, they are leadership questions.

The CEO Mandate

The move from experimentation to transformation is not a task that can be delegated. It is a CEO-level mandate. Capturing the agentic AI advantage requires a fundamental shift in how organizations approach transformation. 

Leaders need to move beyond scattered pilots and align AI with their strategic priorities, treating it as a catalyst for value creation rather than an experiment. This means identifying a few business domains with the highest value potential and pulling every lever: from reimagining workflows to redistributing tasks between humans and machines to rewiring the organization based on new operating models.

Achieving this transformation also demands a new way of working. Siloed AI teams should give way to cross-functional squads that bring together business, data, and technology expertise to deliver impact at scale. Implementation engines should evolve to disciplined, industrial-grade delivery—designed for reliability, scalability, and long-term growth.

The Time for Value Is Now

Agentic AI marks a turning point. The technology is ready. The question is whether leadership will adapt to match it.

Across sectors, CEOs now face a choice: continue experimenting on the periphery or take decisive steps to rewire their organizations around this new paradigm. Those who act with intent — embedding agents deeply into their workflows, investing in governance and data quality, and empowering teams to co-create with AI — will not just incrementally improve performance. They will redefine what their organizations can achieve.