The buzz around agentic AI is undeniable. It's making its way into boardrooms, promising to revolutionize how businesses operate. Yet, as excitement builds, so does confusion. A recent Cloudera survey showed that 96% of global enterprises plan to expand AI agents.

The agentic AI system isn't just an automation layer; it's "the new way to do software." It's a boardroom topic not for its current capabilities, but for its future potential as a key differentiator.

What exactly is Agentic AI?

To separate genuine agentic AI from rebranded automation, we first need a clear definition. For CEOs, an agentic AI system consists of autonomous agents given a task, with the ability to achieve it through multiple actions.

The key lies in the agent itself. An agent, in this context, is an LLM (Large Language Model) running in a loop, with access to various tools and, critically, the ability to make decisions.

For example, asking an AI chatbot for flight options provides a list. A true agent, however, takes an instruction like "book the cheapest flight for me to Bali between 10 AM and 12 PM on July 18th," and delivers a PDF ticket. The crucial difference is the decision-making capability. Many AI agents today simply provide information, leaving decisions to humans. This is due to LLMs being probabilistic, not deterministic models. Over time, as trust in these systems grows, we will see more truly agentic AI in production.

Agentic AI in action: Real-world examples

To understand the practical impact, consider these examples:

• Healthcare Pre-screening: A healthcare provider can use an agent to interact with patients, assess urgency, and direct them to the right doctors and clinics based on various parameters. This automates a complex, human-intensive process.
• E-commerce Marketing: An e-commerce giant employs an agentic system where one agent monitors social media trends to decide campaign themes. Another agent then creates images or videos for these campaigns. After human review and approval, a third agent determines the right audience segment, pushing the ad live. This multi-agent system significantly reduces operational overhead while keeping marketing highly relevant.

These cases highlight the shift from single-task automation to agentic systems where multiple agents collaborate to automate entire processes.

Malaysia's strategic leap towards AI

Malaysian companies enthusiastically embrace agentic AI, seeing it as a chance to redefine operations and gain a competitive edge. Key drivers include addressing the talent gap, achieving operational efficiency to compete globally, and fostering faster innovation.

Malaysian businesses prioritize process reimagination. They understand that simply embedding AI into old processes won't yield a transformative impact. Instead, they rethink processes with AI capabilities in mind, making bold bets backed by solid business cases. This focus on actual ROI positions Malaysia to see significant improvements in operational efficiency, innovation, and global competitiveness over the next few years.

Preparing for the Agentic AI journey

Before diving into agentic AI, leaders must ask two critical questions:

• What is the business's long-term direction? Leaders need a clear vision of their organization in an AI-driven future.
• How can each line of business be reimagined with AI? Empowering department heads to envision future operations with AI drives internal innovation.

This proactive approach is vital. The world is changing, and businesses must adapt by reimagining their current operations.

A foundational element for successful AI implementation is data readiness. Agentic systems rely on access to high-quality, contextualized internal data. Without it, AI initiatives merely act as glorified search engines. Challenges include ensuring data models have business context, addressing data quality issues (often 20-30% of enterprise data is questionable), and clearing existing technology debt. Rushing into AI without a solid data foundation risks accumulating AI technology debt, where new systems quickly become legacy.

Avoiding common pitfalls and embracing the future

Companies often err by pursuing AI for AI's sake, lacking a clear business case or precise KPI mapping. This leads to projects that fail to deliver expected value, causing disillusionment among CXOs. Every AI initiative needs a strong hypothesis, a clear understanding of its impact, and a calculated ROI.

Another common mistake is underestimating the talent transformation required. Many assume current ML or data engineers can simply adapt. However, agentic AI demands a diverse Center of Excellence with practitioners from business, IT, data, and ML teams. While the LLM layer is a commodity, integrating it successfully is a complex software and business problem requiring multi-disciplinary expertise.

When implemented correctly, agentic AI creates an AI-first culture that boosts efficiency across the board. Organizations can see 30% to 80% gains on specific processes, leading to significant financial ROI and a powerful cultural force multiplier that differentiates them long-term.

AI as a whole will be transformative, becoming the "new way to do software." For APAC, this presents both an opportunity to lead and a challenge. While governments focus on upskilling, a talent gap in deep technology AI persists between Western and Asian economies. For APAC to truly be a market leader, it must flip the table, investing in foundational research and driving deep technology transformation rather than simply consuming innovations. This requires bold bets from both governments and enterprise leaders to cultivate a talent pool that can make a lasting regional impact.

Interested in learning more?

Check out this podcast