Ask HN: If 1 person can control 10 AI agents, why would still need that person?

Customers don't actually want to do the work of their vendors. I don't want to scan my own groceries at a supermarket, or check myself in for a flight (if checking backs, or traveling internationally anyway). If I wanted to be doing that kind of work, I'd get a job at a supermarket or an airline.

Because if it takes one full-time person to coordinate the agents, and those agents together perform a function that the client wants to support some other endeavor, in order to replace the agent coordinator, they would have to abandon the other things they do for the endeavor for which they are hiring the firm supplying the agents.

If the client was literally completely passive and had no other role in their task than hiring the firm for something that wasn't part of a larger effort thet were actively engaged in, then, sure, they would have nothing relevant to the task to sacrifice by being the agent coordinator, they'd just need to spend the time learning to be an agent coordinator, and then actually become an active full-time participant rather than completely passive -- but presumably they were passive because that's exactly the level of engagement they were willing to put in, or, at least, because they found that being more active in the task had some associated disutility.

I'm not sure who your clients are. Mine do have goals, but ask me to be creative and come up with the constraints. Literally, I had someone ask me for exactly that on my last call. My folks are nowhere near precise. If they were, they would not need me.

Imagine you are a trader and asked agents to build you a perfect trading system and they discuss every next buy/sell and take a call to execute the action.

How do you know in the long term you are making money? What if they decided to cheat on you and started buying stocks which will definitely crash in 1 month, while showing you a progress of profit by day trading other stocks. Would you just let them do what they're doing?

AI agents still need someone who understands context, prioritizes goals, manages trade-offs, and spots misalignment across outputs. That’s not just orchestration, that’s judgment.

Clients may think they can manage agents directly, but most won’t have the clarity or system thinking to turn AI output into coherent outcomes. That’s where founders, strategists, and builders still add real leverage.

The future isn’t agent vs human it’s high-leverage humans using agents better than anyone else.

So the question should be,

"Will AI agents get good enough, where any individual with just a conversion can get a fully functional AI agent such as a coding agent, a legal review agent, a design agent?"

There's always going to be some level of expertise involved and especially with AI some person has to exercise judgement, watching out for hallucinations.

Patience. People don't pay others to do work they're willing to do themselves.

Some ideas question why we would need "middleman" software at all in a decade: what's the point of Photoshop if you can ask AI to fix blemishes and manipulate images and compose your graphics without it? On some level it's selling picks and shovels after the gold rush.