Every article saying 'THIS IS THE NEW JOB OF THE AI ERA' is one of two jobs. And the first is a lie imo. The first job, is usually something…

Every article saying "THIS IS THE NEW JOB OF THE AI ERA" is one of two jobs. And the first is a lie imo. The first job, is usually something that an AI lab is hiring for and the news blows it completely out of proportion, convincing thousands of people that they should somehow get a certificate for a job that does not exist and pivot into this role. "Prompt engineer" falls into this category - I spoke with one media client in 2023 that was extremely proud of their recent "prompt engineer" hire. My recommendation was to change that person's job immediately. I would argue that "AI storyteller" falls into this category as well, though I honestly just read it as 'really good psychological marketer who has watched Simon Sinek videos'. Everyone needs to be able to communicate to AI (prompt) and everyone needs to be able to communicate to humans (storyteller). Anthropic and OpenAI hire a handful of them and pay them extremely well, but an enterprise like John Deere or Coca-Cola does not need this role today. The second job, is probably a real job, and it's just called a hundred different things. It's AI Operations, AI Operator, Agent Manager, Internal FDE, AI Engineer. These all mean something a little different, some are more technical than others, some lean a bit more toward R&D or office of CIO or COO. Everyone should know how to use agents, but this job is "you are one of the people who really knows how to really use this stuff". If a team wants to build something more complicated than a dashboard, they go to you. If a person is ever going to test a new model, it's you (likely inside some sort of frontier unit). If a team needed to quickly think through token spend and chat best practices, they probably go to you. If a team just "can't seem to get AI to do the thing they want", you parachute in with an eval reset, system instructions overhaul, and multi-agent/provider strategy. It's what we had years ago with an analyst pool - internal, highly trained, and extremely flexible in what department they can drop into and add value to. If I were a new grad, I would try and move into that second job. This person should know how to build agents, agent teams, use both Claude Code and Codex, manage multiple workstreams at once, understand how LLMs work, prompt with /goal, wield loops when it makes financial sense, and have used AI harnesses for 100+ hours. If I were a technical new grad, I would look at AI Engineer or internal FDE (BDE?) type roles. If I were a non-technical new grad, I would look at AI Operations. Share with your friends who still think they should be a prompt engineer.

Source: Allie K. Miller (X) | 2026-06-30