AI Agents and Humans Can Work Together to Maximize Uptime and Reduce Costs

Vijay Jaswal, the Chief Technology Officer for APJ&MEA at IFS, says, open-source models will likely lag behind proprietary Agentic AI systems in the short term due to gaps in orchestration and integration

How do you define Agentic AI, and how does it fundamentally differ from traditional chatbots or rule-based automation?
Very simply put Agentic AI is a system designed to work on its own or autonomously to complete tasks and reach goals without needing step by step instructions. Being the Marvel geek that I am, my favorite analogy to simplify the topic is by using Iron Man! More specifically Tony Stark and his “AI butler” called JARVIS.

JARVIS ‘is’ Agentic AI – It doesn’t just respond. It understands, plans, and acts in a plethora of situations. JARVIS is Tony Stark’s intelligent digital assistant, constantly analyzing data, predicting outcomes, and proactively helping him operate the Iron Man suit and manage his world.

Beyond task automation, what complex, multi-step problems are agentic AIs uniquely positioned to solve that current AI models cannot?
Agentic AI in the real world specifically helps our customers by autonomously solving complex, multi-step challenges across industries. For example, in Enterprise Asset Management, it could continuously monitor asset health, predict failures, and automatically trigger and schedule preventative maintenance. In Field Service, it could dynamically optimize technician routes, reschedule jobs in real-time, and manage spare parts logistics without human input.

For project-based industries like aerospace and construction, it could track costs, detect early risks, and recommend corrective actions to keep projects on time and budget. In manufacturing, Agentic AI could identify production bottlenecks, re-plan operations, and coordinate suppliers to maintain output. It could also analyze the profitability of service contracts, suggest renegotiation or cost controls, and simulate financial impacts. Finally, in sustainability, it could track emissions, spot inefficiencies, and auto-generate ESG compliance reports—acting like a virtual sustainability officer.

How much autonomy should Agentic AI have in decision-making, and where should humans remain in the loop?
Agentic AI should be trusted to run with day-to-day, repeatable, and adaptive tasks, like scheduling, optimization, and issue detection, without constant human input. But when decisions could affect safety, strategy, ethics, compliance, or the bottom line, humans need to stay in control. It’s about letting AI act where it adds speed and efficiency, while keeping people involved where judgment, accountability, and trust still matter most.

What infrastructure challenges (such as power, cooling, latency) arise when deploying Agentic AI at scale?
Deploying Agentic AI at scale brings significant infrastructure challenges, including high power demands due to continuous processing and large model inference, increased cooling requirements in data centers to manage heat from sustained compute loads, and latency issues when real-time decisions need to be made across distributed systems.

Additionally, network bandwidth becomes critical for coordinating multiple AI agents and integrating with enterprise systems, while scalability and uptime must be ensured to prevent disruption to autonomous workflows. These challenges require careful planning across compute, storage, and connectivity layers to support the sustained, autonomous operation of agentic systems.

What’s the next frontier for Agentic AI—will we see AI “agents” collaborating like human teams?
In enterprise environments, collaborative AI agents could manage an entire asset lifecycle: one agent monitoring asset health, another scheduling predictive maintenance, and a third managing spare parts inventory—working together to maximize uptime and reduce costs. In field service, one agent could optimize technician dispatch, another ensures customer communication, and a third handle SLA compliance in real time.

In finance, agents could analyze spending, forecast budgets, and propose reallocation strategies dynamically as business needs change. In the energy sector, teams of AI agents could manage grid balancing, predict demand, and coordinate renewable inputs across regions with minimal manual control. This evolution toward multi-agent collaboration means AI will not just support tasks—it will run processes end-to-end, responding to change, resolving conflicts, and learning continuously as a coordinated digital workforce.

Will open-source models keep pace with proprietary Agentic AI systems, or will there be a widening gap?
Open-source models will likely lag behind proprietary Agentic AI systems in the short term due to gaps in orchestration and integration, but over time they may catch up as communities innovate, tooling matures, and modular agent frameworks become more accessible.