Agentic AI: Reshaping Industries with Collaborative Intelligence
Arturo Buzzalino, Group VP and Chief Innovation Officer at Epicor, sheds light on the transformative power of Agentic AI. He also outlines the fundamental challenges businesses must overcome—including data quality, scalability, and trust—to unlock the full potential of Agentic AI and its future advancements
How do you define Agentic AI, and how does it fundamentally differ from traditional chatbots or rule-based automation?
Agentic AI typically refers to AI systems designed to perform task-specific actions autonomously while collaborating with each other and with users. For example, Epicor Prism, our agentic generative AI service built specifically for the supply chain industries, is based on task-specific AI agents that streamline complex processes and provide actionable insights, differentiating it from passive or purely generative systems.
Beyond task automation, what complex, multi-step problems are agentic AIs uniquely positioned to solve that current AI models cannot?
Unlike more generic AI systems, agentic AI can address industry-specific challenges efficiently, making it particularly advantageous for problem-solving in specialised domains. As a result, agentic AI has the potential to fundamentally reshape how industries operate by reducing friction in decision-making and empowering systems to act with greater autonomy.
In sectors like logistics and manufacturing, for instance, solutions such as Epicor Prisim are helping streamline complex supply chain workflows, from automating routine supplier communications to proactively surfacing the best purchasing options based on real-time data. Currently, a major opportunity is in leveraging Agentic AI to bridge data silos, reduce administrative burden, and support faster, more informed decisions.
How much autonomy should Agentic AI have in decision-making, and where should humans remain in the loop?
Agentic AI gains its decision-making autonomy by blending three pillars: sophisticated learning algorithms, sharp situational awareness, and rapid feedback loops that steer it toward its goals. But its freedom stops at well-defined guardrails — ethical guidelines, company policy, and, ultimately, human oversight. In other words, it can act on its own, but only inside the lanes we’ve painted to keep it aligned with our values.
What’s the next frontier for Agentic AI—will we see AI “agents” collaborating like human teams?
Before exploring the next frontier, I would advise business leaders to get the fundamentals right so that they can achieve what agentic AI systems today are capable of delivering. This means tackling many challenges, perhaps the biggest of which presently is getting the right data. This means not just collecting huge amounts of it, but making sure it’s clean, structured, and useful. Agentic AI needs to understand and react to what’s happening in real time, which means it has to make sense of a lot of moving parts quickly and accurately.
There’s also the question of scale. How do you build something that works just as well for a small business as it does for a global enterprise? And of course there’s trust. If an AI system is going to act on its own, people need to understand how it’s making decisions. That means building in transparency and reliability from the start. These are complex problems, but solving them is key to making agentic AI, and subsequent advancements in the space, truly effective in the real world.
