Agentic AI: Unlocking the Next Frontier of Autonomous Systems
In an era where AI is rapidly evolving beyond simple chatbots, Haider Aziz, General Manager – Middle East, Turkey and Africa (META) at VAST Data, sheds light on the transformative potential of Agentic AI. From its fundamental differences to its unique problem-solving capabilities and the infrastructure demands it presents, we explore how these autonomous systems are poised to reshape enterprise operations and even collaborate like human teams.
How do you define Agentic AI, and how does it fundamentally differ from traditional chatbots or rule-based automation?
Agentic AI is very different to chatbots. Firstly, agentic isn’t just reactive, it’s autonomous. Unlike chatbots or scripted workflows, agents have memory, context, and a sense of purpose. They don’t just answer questions, they make decisions, take actions, and adapt over time based on evolving goals. That requires a complete rethink of how we store, move, and reason over data.
Beyond task automation, what complex, multi-step problems are agentic AIs uniquely positioned to solve that current AI models cannot?
Agents excel at problems that span time, context, and uncertainty, like managing IT incidents end to end, handling multi-channel customer journeys, or navigating dynamic logistics in real time. These are workflows that require planning, adaptation, and continuous reasoning, not just classification or prompt response.
How much autonomy should Agentic AI have in decision-making, and where should humans remain in the loop?
It depends on the stakes. For high-trust, repeatable tasks, agents can and should act independently. But in high-risk or highly regulated domains, human oversight remains essential. What matters is visibility: we need architectures that ensure every decision is explainable, traceable, and reversible if needed.
What infrastructure challenges (such as power, cooling, latency) arise when deploying Agentic AI at scale
Agentic AI demands constant data access, not just bursts of compute. That increases pressure on power, cooling, and especially data locality. As a result, latency becomes a business risk, not just a frustration. The solution isn’t just more GPUs, it’s infrastructure that brings compute to data, reduces movement, and supports always-on reasoning efficiently.
What’s the next frontier for Agentic AI—will we see AI “agents” collaborating like human teams?
Absolutely. We’re already seeing the early stages: retrieval agents, planning agents, and execution agents working in concert with human employees. The real frontier is multi-agent collaboration with shared memory and goals; with agents that specialise, communicate, and self-organise like human teams. That’s where infrastructure truly becomes the nervous system of enterprise AI.
Will open-source models keep pace with proprietary Agentic AI systems, or will there be a widening gap?
Open source will remain a critical force, especially for transparency, local control, and rapid innovation. But building agentic systems isn’t just about the model, it’s about orchestration, memory, and infrastructure. That’s where the gap may widen. Enterprise scale organisations, including government departments will need both: the freedom of open source and the architecture to make it operational.
