.png){kind=logo}
AI Delegation: AI agents can now handle multi-step tasks, so professionals must learn to hand over execution while keeping judgment.
Human Oversight: Decisions that bring money in or take money out should always involve a human.
Future of Work: Free time from AI is likely to go into deeper thinking, planning and review.
Artificial intelligence is moving from simple tools to agents that can plan, decide and complete tasks. These agents handle multi-step work with limited human input. This shift changes how professionals work, which skills matter and who stays accountable.
In this episode of Analytics Insight Conversations, host Priya Dayalani speaks with Dr. Pavankumar Gurazada, Associate Director, AI/Data Science at Great Learning, about the move from AI assistants to AI delegation, human oversight and the future of work.
Could Dr. Pavankumar Gurazada tell us about Great Learning and his role?
I have been with Great Learning for about six years, mostly as an educator. In my current role, I help design and deliver AI and generative AI programs. Most of our learners are working professionals. We also work with large enterprises in India and a few outside the country. My research focuses on applying AI to business problems, so my work sits where applied AI meets the real enterprise world.
How should AI agents change the way professionals think about daily work?
Agents are moving in two directions, personal and enterprise. Personal agents help with tasks like planning a trip. Enterprise agents are more advanced and support knowledge work and business processes. I suggest splitting work into two parts. Recurring tasks, such as reviewing the last 24 hours of emails, can go to an agent. One-off tasks need more human input and should stay with the professional. This is a simple starting point for any team.
Which skills become more important when an agent figures out the steps?
The biggest skill is giving up a little control. Professionals need to be more managerial and less focused on execution. Every person must own the small piece of work they once simply carried out. The thinking and planning stay with the human. The step-by-step tasks go to the agent. The person then owns the outcome and learns from it. This skill comes through experience, and mistakes help people build it over time.
Where is the line between "AI can do this for me" and "AI can recommend, but I approve"?
The default rule is simple. Judgment is never delegated, only execution. A sharp line is money. If a decision brings money in or takes it out, a human should stay involved. Outside that area, agents can often decide from start to finish. It also helps to measure the human baseline first. Human teams are not perfect either, so this comparison gives the business more confidence before it hands work to an agent.
How can professionals review AI output instead of approving it because it looks convincing?
I look at four checks. The first is relevance, which means the output should fit the reader and the goal. The second is information density, so every sentence should carry a fact, number, decision or action. The third is factuality. Models still invent numbers, so professionals should use figures they already know as anchors. The fourth is perspective. Models often avoid taking a stand, and this false neutrality should be removed. These checks are the human's job.
Will professionals work less now that AI can do more of the work?
I do not think so. Time spent on execution has dropped close to zero. Reviewing the output and applying judgment still takes time. Anyone with extra free time is likely not meeting the expectations of the role. That time should go into deeper thinking and upfront planning. This is the harder part of work in enterprises today.