I’m My Boss’s Secret AI Whisperer. It’s the Weirdest Part of My Job.

It usually starts with a ping on Teams. Not in a group chat, always a direct message. “Rohan, have you got a minute?”

I know what it means. It’s not about the quarterly report I filed last week or the upcoming client pitch. It’s a summons to the inner sanctum, my manager’s glass-walled cabin, for a session of what has become the strangest, most unspoken part of my job. I am my boss’s AI whisperer.

He’ll gesture for me to close the door, a conspiratorial air filling the small space. Mr. Sharma is a smart man, a veteran of the industry with two decades of experience. He can read a balance sheet like a poet reads verse. But when he turns his monitor towards me, his face is a mask of frustration. On the screen is a ChatGPT window, and it’s giving him garbage.

“I asked it for a competitive analysis of the Indonesian market,” he’ll say, his voice a low grumble. “Look at this. It’s generic. Useless.”

I’ll lean in and read his prompt. It’s usually something like: ‘Competitive analysis of Indonesian fintech market for our product.’ It’s not a bad start for a human, but it’s a terrible instruction for a machine. It’s like asking a librarian to “find me a book.”

This is where my work begins. I don’t just type for him. I act as a translator. I interview him, teasing out the nuance and context that lives in his head. “What’s our specific product angle, sir? Are we focused on B2B or B2C? Which competitors are you most worried about? What timeframe are we looking at for the analysis? What should the tone be? Formal? A memo for internal leadership?”

He answers my questions, and I begin to craft. My fingers fly across his keyboard, weaving his deep industry knowledge into a tapestry of precise instructions, context-rich commands, and negative constraints. I tell the AI what to be, who its audience is, what to include, and, just as importantly, what to ignore. It’s a dialogue. I’m not just prompting; I’m shaping, guiding, and refining the machine’s thought process. I’m the ghostwriter for a conversation between my boss’s brain and a silicon one.

Within ten minutes, we have it. A crisp, insightful analysis, complete with tables and specific data points it cross-referenced from its training data. It’s not perfect, but it’s an 80% solution that saves him a day’s worth of work. He’ll look at the screen, a slow smile spreading across his face. “Brilliant, Rohan. Absolutely brilliant.”

Then comes the most surreal part of this whole ritual. He copies the text, pastes it into a Word document, and then deletes the entire AI chat history. As if it never happened. I’m dismissed with a grateful nod, and I walk back to my desk in the open-plan office, our little secret dissolving into the air between our workstations.

This dynamic didn’t start overnight. It began subtly about a year ago. The mandate came from on high: “Leverage AI to improve efficiency.” For digitally native employees like me, it was an exciting invitation. I had been tinkering with large language models for months, using them to synthesise research, brainstorm ideas, and clean up my code. Mr. Sharma, however, was hesitant. He saw it as a toy, or worse, a threat.

One afternoon, he saw me use it to generate a dozen different taglines for a marketing campaign in about 30 seconds. He was stunned. He called me into his office, the first of many such secret meetings. “Show me how you did that,” he’d asked. I showed him. The next week, it was, “Can you help me with this?” And now, it’s often just an implicit understanding. He has the need; I have the skill.

I have friends in Bangalore, London, and San Francisco. This isn’t just happening in my Mumbai office. It’s a silent, global phenomenon. A whole generation of senior managers, excellent at their jobs, are now faced with a technology that feels both alien and essential. They are caught between the pressure to adopt it and the fear of looking incompetent. So they find people like me. We are the unofficial, unlisted, unpaid AI consultants for our own bosses.

In India, this dynamic is layered with our unique culture of hierarchy. You don’t tell your boss, a man twenty years your senior whom you call ‘sir’, that his prompts are lazy. You don’t suggest he take an online course. That would be disrespectful, an affront to the established order. So you step in, quietly, and become the hidden engine. You allow him to maintain his authority while you provide the modern horsepower.

Last month was the peak of this weirdness. We had a major presentation for the Asia-Pacific leadership team. For a week, Mr. Sharma and I were practically attached at the hip. We spent hours after work in his office, him dictating strategy and vision, me translating it into a series of mega-prompts. We built the narrative, the talking points, the market projections, and even the Q&A prep guide, all through an iterative process with the AI.

Then came the presentation. I sat in the third row of the boardroom as Mr. Sharma, confident and eloquent, walked the executive team through the slides. He presented the AI-generated five-year growth model as his own strategic foresight. He used the AI-scripted phrases to answer tough questions with ease. He received a round of applause. The regional head commended him on his “innovative thinking.”

I felt a bizarre cocktail of emotions. There was pride, because our work—my work—was being praised. There was a twinge of resentment, a feeling of being a ghostwriter at my own play. But mostly, there was just profound awkwardness. I had to clap along, to smile and nod at the brilliance of the man who was, in a way, performing a script I had co-written with a machine.

It makes you question things. Who is the impostor here? Is it him, for presenting work he couldn’t have produced on his own? Or is it me? Am I just a glorified typist, a human API call? Does this secret skill even count as real expertise? It’s not on my job description. It won’t be on my performance review. It’s a shadow skill, valuable only in the shadows.

This clandestine arrangement inverts the power dynamic in a fascinating way. In that closed room, he is dependent on me. His immediate success hinges on my ability to 'speak machine'. But the moment I step out, the traditional hierarchy snaps back into place. I am his junior report; he signs off on my leave requests. It’s a form of job security, I suppose, but a fragile and unacknowledged one. If he ever learns to do it himself, or if the technology gets so simple that anyone can do it, my secret value evaporates.

I wonder if I’m doing him a disservice. By being his crutch, I am preventing him from learning a fundamental skill for the next decade of business. I’m helping him survive the present, but I might be hindering his ability to thrive in the future. I’m a temporary fix, a human patch for a systemic skills gap.

For now, the pings on Teams continue. The closed-door sessions go on. I’ll continue to be the ghost in his machine, the whisperer to his AI. It’s the weirdest part of my job, and maybe the most quintessentially modern. I am a bridge between two generations, two modes of thinking, two eras of work. And standing on a bridge is a very strange, very precarious, and very interesting place to be.