EVOLVING faster that humans can catch up to, artificial intelligence (AI), has moved from just being foundational to learning to talk to us in the large language model (LLM) to becoming generative in its reasoning and then agentic, meaning autonomous.
Agentic AI refers to machine learning agents that can mimic human-like actions to execute complex tasks, embodying four central elements: planning, decision-making, task execution orchestration, and human feedback. These sophisticated systems are designed to simplify processes, enhance user experiences, and maximize efficiency across various industries, with practical applications resonating globally, including within the Philippines.
IBM has been at the forefront, driving advancements in areas like Agentic AI.
Malaya Business Insight had a chance to interview Kitman Cheung, a seasoned AI expert—an engineer, software developer, author and an inventor who has worked across different domains like Data, AI, DevOps, Hybrid Cloud, Business Automation, Cyber-Security and AIOps. He also has a knack for breaking the technicalities of technology into relatable, bite-sized pieces of information.
During the interview, he dissected the core principles of Agentic AI. According to Cheung, these AI systems can plan, make decisions based on contextual guidelines, execute tasks across multiple systems, and learn from outcomes, all while incorporating human verification to ensure reliability and safety. He likened the development to the transition from human-assisted travel planning to self-service platforms, lamenting the loss of convenience that came with the latter.
Cheung emphasized that Agentic AI’s practicality lies in its ability to engage in complex, nuanced tasks that require not just execution but also strategic planning, decision-making, and adaptability to new information or changing circumstances. The agent effectively acts as an extension of the user, orchestrating intricate workflows and making informed decisions within predefined parameters, all while accommodating human feedback and intervention.
A notable concern in the realm of advanced AI is cybersecurity and data protection. With increased accessibility and complexity, AI systems open up additional attack vectors for malicious actors. Cheung addressed this by highlighting the importance of model transparency, curation of training data, and robust governance frameworks at IBM.
IBM emphasizes the development of models in an open and transparent manner, with training data curated to remove sensitive or harmful content. This approach is crucial for building trustworthy AI that operates within corporate environments without compromising security. Additionally, employing governance frameworks for continuous monitoring of inputs and outputs, along with predefined guardrails, prevents AI models from generating inappropriate or damaging outputs.
IBM has demonstrated the potential of Agentic AI through practical implementations tailored to specific industries. Cheung discussed several examples, including the AI-driven HR chatbot within IBM itself, which assists with tasks ranging from leaves approval to onboarding. The success of such implementations stems from considering the business problem holistically and selecting the right technology to address the identified pain points, rather than forcing a singular AI solution into diverse operations.
Cheung also extended the conversation to potential applications in supply chain management, where Agentic AI could streamline complex processes, enhance traceability, and improve responsiveness to disruptions. This is particularly relevant for the Philippines, a country rich in manufacturing and export industries, where efficient supply chain management significantly impacts economic performance.
The prospect of Agentic AI taking root in the Philippines is compelling, given the nation’s aspiration for digital transformation and its strategic location in a rapidly evolving Southeast Asian market.
One practical application could be in the public sector, where Agentic AI could optimize government services, making them more accessible and user-centric. For example, an AI agent could assist citizens in navigating complex bureaucratic processes, such as business registrations or permit applications, by breaking them down into manageable steps, providing accurate real-time information, and guiding citizens through the required documentation and forms.
As President Ferdinand Marcos Jr.’s administration zeroes in on improving digital infrastructure and enhancing the ease of doing business, adopting advanced AI technologies could be instrumental.
In the private sector, especially within the expansive BPO industry, Agentic AI could revolutionize customer service by providing intelligent and context-aware virtual assistants capable of handling inquiries across various platforms. These AI agents could learn from interactions, improving their ability to resolve issues and provide personalized support, ultimately enhancing the Philippines’ reputation for excellence in customer service outsourcing.
Moreover, considering the country’s rich biodiversity and growing interest in ecotourism, Agentic AI could also play a significant role in creating personalized travel experiences for tourists. By integrating with existing travel platforms, these AI agents could plan vacations, consider personal preferences, and adapt to real-time factors like weather conditions, offering travelers a seamless and tailored adventure.
Will Agentic AI bring forth an exciting future where efficiency, personalization, and security converge? The key, as IBM has shown, lies in human-centered design and tailored solutions that address specific pain points within organizations, respecting the unique character of each business context.
