In this interview, Mostafa Kabel, General Manager, Solutions & Services, Mindware, explains the transformative power of Agentic AI. He describes how these intelligent systems are moving beyond traditional automation to actively pursue goals and make independent decisions, fundamentally reshaping how businesses operate.
How do you define Agentic AI, and how does it fundamentally differ from traditional chatbots or rule-based automation?
As artificial intelligence continues to evolve, we are entering a new era defined not by static automation, but by dynamic, autonomous intelligence. This new wave is called Agentic AI — systems that don’t just respond to commands, but actively pursue goals, adapt strategies, and make independent decisions. At Mindware, we see Agentic AI as a defining step toward a future where machines can manage complex tasks, collaborate like teams, and reshape how enterprises operate.
Traditional chatbots and rule-based systems are like scripts — they follow pre-defined paths. They’re useful for basic Q&A, form processing, or linear workflows, but they fall apart when the path isn’t clear or when flexibility is needed. Agentic AI is fundamentally different. These systems combine planning, memory, reasoning, and adaptability. They understand the context, assess multiple options, learn from past actions, and self-correct in real time. Rather than executing tasks, they achieve objectives. This is a leap from reaction to intention.
Beyond task automation, what complex, multi-step problems are agentic AIs uniquely positioned to solve that current AI models cannot?
Agentic AI unlocks a new class of enterprise capabilities. Unlike conventional AI, which focuses on narrow tasks, Agentic systems can operate across multiple stages and dynamically shift their course. Examples include autonomous IT operations, where systems identify, diagnose, and resolve incidents end-to-end without human intervention.
In supply chain optimisation, Agentic AI can react to delays or disruptions in real time by re-routing resources and adjusting logistics plans autonomously. For customer experience management, these systems go beyond responding to complaints—they proactively identify friction points, personalise responses, and coordinate resolution across departments. These are not workflows you can script. They demand adaptive, goal-oriented intelligence—the hallmark of Agentic AI.
How much autonomy should Agentic AI have in decision-making, and where should humans remain in the loop?
One of the most critical debates in Agentic AI is: how much autonomy is too much? In my view, context is key. For routine, low-risk activities, full autonomy increases efficiency and removes bottlenecks. But when it comes to ethically sensitive or high-impact decisions, humans must remain in the loop. An ideal model is “AI proposes, human approves.” Agentic AI can generate solutions, forecast risks, and manage execution — but strategic decisions, accountability, and values must remain human-driven.
What infrastructure challenges (such as power, cooling, latency) arise when deploying Agentic AI at scale?
While the potential is massive, deploying Agentic AI at scale is no small feat. These systems place significant demands on IT infrastructure. Compute power is critical, as goal-driven agents require persistent compute cycles, often leveraging GPU acceleration and high-memory architectures. Energy and cooling needs increase significantly as workloads intensify, especially for concurrent multi-agent systems.
Latency and edge readiness are also essential, particularly for real-time responsiveness in use cases like industrial automation and smart retail. Scalability presents another challenge, as Agentic systems aren’t one-model-fits-all—you need infrastructure that supports simultaneous, collaborative agents across regions. At Mindware, we’re investing in composable infrastructure solutions and partnerships with all the top AI vendors to meet these demands while staying flexible and scalable.
What’s the next frontier for Agentic AI—will we see AI “agents” collaborating like human teams?
The most exciting frontier of Agentic AI is the emergence of multi-agent collaboration. Just as human teams bring different skills to solve problems, we will soon see AI agents working together—one researching, another planning, another executing. This transforms industries. In marketing, agents could run entire campaigns, with one writing copy, another targeting segments, and a third measuring results. In engineering, AI agents might co-design products, test simulations, and optimise supply chains in continuous feedback loops. The vision is no longer science fiction. It’s engineering in motion.
Will open-source models keep pace with proprietary Agentic AI systems, or will there be a widening gap?
There’s a growing debate around whether open-source AI can keep pace with proprietary platforms in this space.
Open-source has democratised access and accelerated experimentation. But Agentic AI is more than just model size — it’s about orchestration, safety layers, memory persistence, and multi-agent communication. Proprietary platforms still lead in these areas. That said, hybrid models will likely emerge — open-source cores, extended by proprietary orchestration layers. The race will not just be about capability, but about trust, interoperability, and control.
