In recent years, businesses have increasingly utilized artificial intelligence (AI) to assist with structured and repetitive tasks. However, the emergence of agentic AI marks a significant shift in this trend. This transition is akin to comparing a GPS that provides route suggestions to a self-driving car that autonomously navigates to a destination. The evolution from passive, analytical systems to proactive, autonomous agents capable of understanding goals and executing complex, multi-step tasks represents a critical advancement, all while adhering to established guidelines.
According to the NASSCOM AI Adoption Index, 87% of Indian enterprises are actively implementing AI solutions, predominantly in customer-facing roles to enhance user experience and optimize marketing strategies. While these applications are significant, the true transformative potential lies in AI’s ability to drive end-to-end processes. In this new paradigm, AI agents assume the role of frontline workers, executing complex tasks and allowing human counterparts to focus on governance, strategic planning, and overarching objectives.
Consider the full journey of a single customer order. Traditionally, this process involves multiple human interactions, including order entry, inventory verification, logistics coordination, and invoice management. An agentic AI system transcends these limitations by autonomously placing purchase orders with suppliers based on predefined rules, tracking shipments, and updating inventory management systems upon receipt, all while keeping relevant teams informed. This system evolves from a mere advisor to an active participant in operational workflows, capable of managing more ambiguous scenarios that previously required human judgment. In the realm of accounts payable, for instance, AI has progressed from manual data entry to rule-based automation, and now stands on the brink of an autonomous future. In this future, AI agents can learn a supplier’s invoicing patterns, manage approvals, adapt to changes, and optimize payment schedules, needing human intervention primarily for exceptions.
The concept of autonomous AI agents raises concerns regarding potential rogue operations. Such apprehensions are valid and can be addressed through a combination of robust technical architecture and human oversight. Visualizing this relationship can be enriched by the ‘cop on the beat’ analogy. AI agents, analogous to police officers, operate within clearly defined boundaries to manage routine tasks, while more complex decisions requiring nuanced judgment are best handled by human ‘special agents.’ In enterprises, full autonomy should never equate to a lack of supervision.
Another significant challenge lies in enterprise debt, an entrenched complexity of outdated processes and inconsistent data afflicting nearly all large organizations. Building effective, intelligent agents necessitates first addressing this foundational debt. A comprehensive approach to training AI agents is critical; it is inadequate to merely supply them with data. Real intelligence emerges from the integration of three essential knowledge pillars: data, processes, and tacit knowledge. Data pertains to transactional history and organizational information. Process knowledge encompasses standard operating procedures and defined workflows. Tacit knowledge represents the unwritten expertise of experienced employees, which is often the most neglected aspect. This intuitive, organizational wisdom empowers teams to navigate the complexities of challenging scenarios where human involvement is essential.
By interweaving these three elements, organizations can develop AI agents that not only adhere to established rules but also possess the capacity to problem-solve and improvise in alignment with business operations.
India is well-positioned to capitalize on these developments in agentic AI, thanks to its expansive and growing talent pool. Discussion in the Indian technology sector is shifting from experimentation to widespread application of AI solutions. With a substantial percentage of organizations exploring agentic AI, there is a notable eagerness for this new wave of automation. The Indian government’s initiatives, such as the India AI Mission, are expected to further propel this momentum by providing essential infrastructure and support for innovation. The current challenge for Indian enterprises is to move beyond isolated proofs of concept to strategically embed AI into core business processes for measurable outcomes.
The author is Vijay Vijayasankar, Global Agentic AI Officer at Genpact.
Disclaimer: The perspectives presented are solely those of the author, and ETCIO does not necessarily endorse them. ETCIO assumes no responsibility for any potential harm incurred by individuals or organizations, whether directly or indirectly.
