The 18-month Career Is Here

Was this newsletter forwarded to you? Sign up to get it in your inbox.

If anyone is a prompt engineer, Michael Taylor is. He wrote an essential book on it. He writes an Every column about it. He's building Ask Rally, an AI startup that creates synthetic market research by modeling your target audiences—and he says prompt engineering is “most of the work.”

He also thinks he has about five years before the job disappears.

If the headlines are to be believed, five years may be too generous. Less than two years ago, the Wall Street Journal wrote about prompt engineering roles paying up to $200,000. By this spring, the same publication declared the job "already obsolete."

This is the whiplash of careers in the AI economy: entire disciplines rising and falling and getting folded into the knowledge worker skill stack faster than you can update your LinkedIn profile. Speaking as someone who recently swan-dived into a new job as an "AI operations lead," I'd be lying if I said it doesn't freak me out. If it's prompt engineering today, will it be AI operations tomorrow?

The answer is: Maybe. And I’m learning to be OK with that. The path started crumbling long before AI—think pandemic pivots, the freelance boom, the rise of portfolio careers. What AI does is accelerate the shift so dramatically that ignoring it becomes impossible. And when the ground moves this fast, moving with it might be where the opportunity is.

The brief, wondrous life of a bridge job

Mike's five-year plan makes more sense when you understand what prompt engineering really is: a bridge job. These roles show up when a new technology needs human translators. They’re essential at first, then absorbed once the interface gets easier.

In the 1950s, companies hired entire pools of typists. By the nineties, everyone typed their own emails. Webmasters were a hot commodity in the early internet days. Now, your marketing intern spins up a site on Squarespace in 20 minutes.

The difference now is speed. Typists had decades. Webmasters had years. With prompt engineering, the whole arc compressed into 18 months. The shelf life of a bridge job used to buy you a career. Now it barely covers onboarding.

I’ve seen it up close. For the past few weeks, I’ve been working on a custom AI proofreader for Every—a tool to catch grammar and mechanics issues before a human review. Last week, Dan Shipper, our CEO, added our style guide as a skill in Dia, the AI-powered browser. For now, that means Dan doesn’t need my AI proofreader to write Every-compliant copy. Today it’s Dan; tomorrow—if he were to add that skill to Dia’s skill library for other Dia users to access—it could be everyone.

That’s what it’s like to work in AI. You build tools for problems that might not exist in a quarter. Every morning, I check what models improved overnight, what workflows my coworkers built, and which of my systems have already been made irrelevant.

It’s a pattern I expect we’ll see more of as model sophistication accelerates: not just individual tools getting replaced, but entire skill sets evolving in real time. We’re not moving vertically up ladders, or even horizontally on jungle gyms. We’re in open territory, drawing the map as we go.

Build your product, outsource the marketing page

You've got bigger things to worry about than website copy and dev timelines. Framer lets you vibe code marketing pages that actually represent your brand, with scroll animations, custom interactions, and conversion tools built-in. Focus on the work that matters.

Use code EVERY2025 for a free month of Framer Pro

Want to sponsor Every? Click here.

What comes after obsolescence

The traditional career model assumes skills and roles remain stable long enough to build expertise and advance within them. AI breaks that assumption. When job titles can emerge and disappear within 18 months, building identity around specific technical skills becomes risky.

The sustainable approach might be Mike’s: Focus on problems that persist even as tools evolve. His work extracting insights from complex datasets will remain valuable whether it requires prompt engineering, fine-tuning models, or techniques that don't exist yet.

My own experience suggests three principles for navigating this:

Follow the friction. The work that feels tedious to you probably feels tedious to others, too. When you solve it, you're solving a real problem that creates value. The AI editor emerged from a core Every principle: Don't repeat yourself. If we're catching the same formatting issues weekly, that's friction worth automating.

Build taste, not just tools. Anyone can learn to use Claude or ChatGPT. Fewer people develop the judgment to know when the output is good, or how to make it better. That discernment becomes more valuable as the models get more sophisticated: The more context they understand, the more important giving them the right context is.

Bring your baggage. Mike brought marketing instincts to AI. I brought editorial standards to operations. The expertise you think is irrelevant—your "non-technical" background—might be exactly what makes you valuable. Domain knowledge doesn't expire as fast as technical skills.

The view from a burning bridge

The $200,000 prompt engineering job is gone. But the skill isn't. Just this week, Michael was posting to X about research showing that prompt optimization is more important than fine tuning for delivering results. Two years ago, companies threw six figures at the job. Six months ago, they declared it dead. Now the skill may be more valuable than the "real" engineering everyone said would replace it.

Your specialty—whatever it is—may be somewhere on this arc. Mike's betting on it with his five-year timeline. But watching him optimize prompts that engineers swore would be irrelevant, watching my own job description shift weekly, I'm realizing this constant change creates its own kind of advantage.

Your skill becomes baseline, sure. But that also means you're the first to spot the gaps. Instead of mourning the bridge you're on, recognize that you have the perfect vantage point for spotting where the next one should go. And if you see it first, you get to help design it.

Katie Parrott is a writer, editor, and content marketer focused on the intersection of technology, work, and culture. You can read more of her work in her newsletter.

To read more essays like this, subscribe to Every, and follow us on X at @every and on LinkedIn.

We build AI tools for readers like you. Automate repeat writing with Spiral. Organize files automatically with Sparkle. Deliver yourself from email with Cora.

We also do AI training, adoption, and innovation for companies. Work with us to bring AI into your organization.

Get paid for sharing Every with your friends. Join our referral program.