Let's pour one out for the dumbest gold rush of 2024: the "Prompt Pack."
You remember them. "1,000 Ultimate ChatGPT Prompts for Real Estate Agents." "The $99 Prompt Bible for Marketers." Pitched by gurus, sold on Gumroad, and never used by anyone, ever.
It was a grift. A low-effort, cynical cash grab selling people something they didn't need.
Why did it fail? Because a business owner doesn't want a PDF of 1,000 instructions. They don't want to learn "prompt engineering." They just want their damn problem to go away.
The prompt pack is dead. Good riddance.
But the problem is bigger than ever. Businesses are drowning in AI hype. They know this tech is powerful, but they are stuck, getting garbage results from garbage inputs. They are desperate for a translator.
That's the real, temporary, high-ticket opportunity.
Stop Selling Instructions. Start Selling Integration.
You're not a "Prompt Engineer." That's a fake title for a skill that won't exist in five years.
You are an AI Workflow Consultant.
You don't sell prompts. You sell systems. You don't sell instructions. You sell integration.
Your client doesn't care about the perfect prompt. They care about automating the 10 hours a week they spend answering the same five customer service emails. They care about a system that drafts their 20 property descriptions. They care about turning a 50-page report into a 1-page summary without them having to do it.
You are the one who builds that system.
What You Actually Sell
This is a consulting business. You're not selling $49 PDFs. You're selling $2,000 projects.
- The AI Audit ($1,000): You go into a business for a week. You analyze their workflows. You interview their team. You identify the 5 biggest time-wasting, automatable tasks. You deliver a 3-page "Automation Roadmap."
- The Implementation Project ($3,000 - $5,000): You take that roadmap and you build it. You create the custom prompt frameworks for their specific needs. You build the templates. You train their team on how to use them. You are the one who makes the AI actually work for their business.
- The Optimization Retainer ($1,500/month): You stay on to refine the systems, answer questions, and build new workflows as new problems arise.
This isn't about "hacks." It's about deep, valuable, operational work.
This Is an Arbitrage Opportunity. Not a Career.
Now, here's the part no one wants to say out loud: this job is temporary.
This is a classic arbitrage opportunity. You are capitalizing on a temporary gap in the market: the gap between what AI can do and what the average person knows how to do.
That gap is closing.
In 2-3 years, AI tools will be 10x better at understanding vague, human intent. This "skill" of prompt engineering will become as common (and as un-monetizable) as knowing how to use a Google search bar.
This isn't a 30-year career. It's a 3-year smash-and-grab.
You're not building a "prompt engineering agency." You're building a bankroll. You are capitalizing on a transitional moment in technology.
So bill accordingly. Charge premium rates. Take the money. Because in 2028, this job won't exist. And you'll be on to the next thing.
Scoring the prediction: August 2026
Editor's note. This essay made a dated, falsifiable prediction, so it deserves a dated scoring. Original October 2025 text unchanged above.
The essay said: prompt engineering as a title is a temporary arbitrage, the gap will close, and by 2028 this job won't exist. We're roughly halfway to that date. Here's the scorecard.
Right about the title. The standalone "Prompt Engineer" job posting has been shrinking, not growing — hiring analyses through 2026 put the decline in the title at roughly a third from its 2024 peak. It never became a profession. It was a job posting fashion, and the fashion passed.
Wrong about the work. This is the part the essay got backwards. The skill did not become un-monetizable; it got absorbed. Postings that require prompting ability as one competence among several have multiplied over the same period, folded into AI engineering, applied science and ordinary product roles. The gap didn't close so much as it moved: what's scarce in 2026 isn't people who can write a good prompt, it's people who can evaluate whether the output is right and ship the thing around it.
Still open. Whether the consulting business described here — audit, implementation, retainer — survives to 2028 is genuinely unresolved. The audit is holding up. The "build custom prompt frameworks" implementation work is thinner than it was, because the tools got better at vague human intent exactly as predicted.
If you took this essay's advice in October 2025 and billed accordingly, you did fine. If you took the "3-year smash-and-grab" framing literally and skipped building anything durable underneath it, the exit is arriving earlier than 2028.