The bottom rung is gone

She's not applying wrong. Her resume isn't broken.

She's tailoring her applications, hitting "Easy Apply," going through the referral channels her professors laid out. She's targeting content strategy and marketing analytics roles — entry-level stuff, the kind that existed three years ago in abundance.

She's sent over 40 applications. A handful of interviews. One offer came through in October — then got pulled two weeks later when the company announced a hiring freeze. The rejection emails all say "we went with someone with more experience." The rest say nothing at all.

74% of graduates in a 2025 Stepstone study reported being ghosted entirely — no response, no rejection, just silence. Jordan needed 40 applications to get one interview. The median for a grad her age.

She wasn't failing. She was encountering something nobody had named for her yet.

Here's what the data says happened — and why it happened to her sector first.

⚡ ⚡ ⚡

📊 Three numbers. One pattern.

A Stanford Digital Economy Lab and ADP payroll study — covering 10 million workers — found that employment for workers aged 22–25 fell 13% in highly AI-exposed industries between 2022 and 2025. Workers over 30 in the same sectors? Flat or growing.

The companies didn't downsize. They just stopped backfilling the bottom.

A Stepstone Group analysis of 4.6 million job ads found that entry-level postings in Q1 2025 were 45% below the five-year average — lower than during the first months of the pandemic. This isn't a slow hiring market. It's a structural withdrawal from a specific rung of the ladder.

Thirty-seven percent of companies plan to replace entry-level roles with AI. Fifty-eight percent plan to automate operations and back-office work. Both pools are where careers got built.

To be fair — and this matters: AI tools can accelerate novice performance. MIT researchers found that junior workers gain access to expert playbooks they'd have taken years to build on their own. That's real, and I don't want to pretend it isn't.

But it assumes you're inside the job where those tools are deployed. Jordan can't access the apprenticeship accelerator if nobody will hire her into the apprenticeship.

Here's the part that breaks the "just learn AI" advice: the new roles being created expect the judgment of someone who's been doing the work for five years — in a technology that's been mainstream for three. You can't retrain into that. You either already have domain experience, or you're looking at a ladder with no bottom rung.

Aneesh Raman, who runs workforce strategy at LinkedIn, called it directly: the bottom rung of the career ladder is breaking. He's right. What he hasn't said loudly enough is what that means for the generation that was supposed to be standing on it.

This isn't disruption. It's a trap with a P&L rationale.

But here's what nobody in charge is saying — what actually survives.

🔥 🔥 🔥

It just surpassed a $1B valuation, joining private US companies like SpaceX and OpenAI. Unlike those companies, you can invest in EnergyX today. Industry giants like General Motors and POSCO already have. Why? EnergyX’s tech can recover 3X more lithium than traditional methods. Now, they’re preparing 100,000+ acres of lithium-rich Chilean land for commercial production. Buy private EnergyX shares alongside 40k+ people at $11/share through 2/26.

Invest in EnergyX Today

_{This is a paid advertisement for EnergyX Regulation A offering. Please read the offering circular at invest.energyx.com. Under Regulation A, a company may change its share price by up to 20% without requalifying the offering with the Securities and Exchange Commission.}

🛠️ The thing that isn't going anywhere is judgment. The problem is that AI ate the curriculum.

The roles that are holding — and in some sectors growing — have something in common. They require the ability to make calls that can't be scripted, to contextualize information in ways a model can't verify, to read a room or a client or a situation and decide.

That's not a skill you learn in a bootcamp. It's built through repetition in the field. The exact repetitions AI just automated away.

Trades are a useful contrast. ServiceTitan's 2025 industry report found that nearly half of contractors are already using AI and reporting efficiency gains — but the wrench still belongs to the plumber. Separately, a Gulfshore Air HVAC company integrated AI into scheduling and dispatch and saw 53% revenue growth, while their technicians kept every client-facing job. The physical, the relational, the contextual — these have more runway than most white-collar entry-level work.

Inside white-collar roles, the workers who are outperforming aren't the ones who "learned AI." They're the ones who already had a domain and layered AI underneath it. A marketing manager with eight years of client strategy experience using Claude to triple her output is not the same situation as a 24-year-old being told to learn Claude before she's ever run a campaign. The tool is the same. The underlying asset is completely different.

That's the real gap. Not AI fluency. Domain depth. Which brings me to the actual opportunity nobody's naming.

⚡ ⚡ ⚡

Jordan's problem isn't a skills gap. It's an experience gap — and the conventional path to filling it just closed.

Most of the advice aimed at workers right now assumes you need to acquire something new. Learn a tool. Finish a certification. Master a workflow. That advice is designed for a world where the skills gap is the issue. It isn't.

But here's what Jordan actually has: eight months of job search data, a degree's worth of coursework in consumer behavior and analytics, internship experience, and a working understanding of how marketing campaigns are structured.

She's never run one professionally. She's never owned a budget or managed a client. But she has more domain-adjacent knowledge than she's been told to count.

The pivot isn't "become an AI expert." It's "map what you already know onto where AI needs human oversight." Where does a model get it wrong in your domain? Where does contextual judgment matter more than output volume?

Here's what that looks like in practice. Noel spent 12 years in content marketing — writing, editing, building teams. In December 2024, his company was wiped out overnight by Google's algorithm changes. He was the sole earner for his family with 18-month-old twins. He applied to 60+ jobs. Zero offers.

Then he did something different: instead of learning a new skill, he audited the skills he already had. Twelve years of knowing which content builds trust and which sounds like a machine wrote it. Two weeks after filing his LLC, he had his first clients — offering content audits, quality assurance, and AI editing for companies who discovered their AI output needed a human who actually knew the field. (how he describes the pivot)

That's not a story about someone who got lucky. It's a story about someone who stopped asking "what do I need to learn" and started asking "where does this technology still need a human who actually knows this field."

Those aren't job postings yet — not in any standardized form. But they're the shape of where defensible roles are forming. That's exactly what the Job Audit Prompt below helps you map. Not what you need to learn. What you already have. Zero cost. Ten minutes tonight.

The pivot isn't "become an AI expert." If you want to do this tonight, map what you already have in 10 minutes — the full prompt is one click away.

🔥 🔥 🔥

Most "AI vulnerability" tools ask the wrong question. They analyze your job title against automation probability scores and spit out a percentage. Useful for op-eds. Not useful for you.

This prompt does something different. It asks Claude or ChatGPT to act as a labor economist and map your specific experience, not your job title, against where human judgment is actually holding value.

Here's the core of it — paste this into Claude or ChatGPT tonight. (Don't worry about the "labor economist" part — it just tells the AI how to think. You describe your job in plain English.)

That's the tease. The full prompt — including the follow-up questions that get into specifics by industry and role type — is in the deep-dive email.

Or reply "Ladder" to get it directly. I read every one.

In 2022, Klarna laid off approximately 700 employees, publicly crediting its AI assistant with handling the equivalent customer service workload. CEO Sebastian Siemiatkowski called it a milestone. Analysts wrote it up as the future arriving ahead of schedule.

By 2024–2025, Klarna was quietly rehiring humans into customer-support roles. Customer satisfaction had fallen. Siemiatkowski acknowledged they'd gone "too far" and that human workers remained essential in ways the initial bet hadn't fully accounted for.

The lesson isn't that AI doesn't work. It's this: companies will make this bet with your job before the answer is in. Klarna's employees bore the cost of the experiment. Some got their jobs back. Many didn't. The P&L recovered. Their careers didn't pause while it did.

Jordan isn't just competing with AI. She's also competing with the aftermath of AI decisions that didn't pan out — and the laid-off workers those experiments put back into the same applicant pool.

If you read this far — you're ready to do something with it. Run the audit yourself — it takes 10 minutes.