When Junior Hiring Dies, Your Expertise Becomes the Last Moat

Companies stopped training juniors because they're betting LLMs will absorb that tier. The apprenticeship pipeline is collapsing. Your IP is the last unextracted asset.

Your junior pipeline is gone. Not slowing. Gone.

Entry-level hiring at US professional services firms fell over 30% between 2023 and early 2026. The replacement cost of that vanished tier — the salaries you're not paying — looks like savings on a P&L. It isn't. It's a $2-4M knowledge debt accruing inside every mid-sized expert firm right now, and the bill comes due the moment a senior partner retires, leaves, or sells.

The Financial Times blamed remote work. That's the comfortable answer. The real answer is that CFOs ran the math on Claude and GPT-5 and decided the bottom rung of the org chart was a line item, not a pipeline. They're betting LLMs absorb that tier entirely.

They're half right. And the other half is where the opportunity lives.

I'm Matt Cretzman. I build systems that turn expert knowledge into owned, monetizable IP before someone else extracts it for free. Here's what the death of junior hiring actually means for you, and the move most experts are missing.

The Apprenticeship Pipeline Was a Knowledge Transfer System

For 150 years, professional services ran on a simple deal. Juniors traded cheap labor for proximity. Seniors traded time-on-task for a successor. The firm got continuity. Everyone won, slowly.

That pipeline was never really about labor. It was the transfer mechanism for tacit knowledge — the stuff that lives in a senior's head and shows up as a raised eyebrow during a client meeting. You learned it by watching. By getting corrected. By being in the room when the deal almost died and someone with twenty years of scar tissue knew exactly which sentence to say next.

You can't put that in a memo. You couldn't, anyway.

Now companies are running an experiment. Skip the junior tier. Let the LLM do the grunt work — the first-draft memo, the doc review, the model build. Keep the seniors. Save the salaries.

The experiment works for about thirty-six months. Then the seniors start retiring with nobody trained to replace them, and the firm discovers it has been eating its own seed corn.

Three Things Are Happening at Once

First: the LLMs already ate the public version of expert knowledge. Every textbook, every published paper, every blog post, every Stack Overflow answer is in the training data. The base layer of any profession — the stuff a second-year associate used to learn from a treatise — is now ambient. Free. Available to anyone with a $20 subscription.

Second: platforms like Mercor are paying experts $100-$300 an hour to dump the next layer — the judgment layer, the pattern-recognition layer — directly into training sets. The Mercor CEO has been candid about this on podcasts. They are buying the tacit knowledge of doctors, lawyers, engineers, and financial analysts by the hour and feeding it to frontier labs. The expert gets a 1099. The lab gets the moat.

This is Jacob selling his birthright for a bowl of stew. Hourly. Voluntarily. With a tax form.

Third: because juniors aren't being hired, the in-firm transfer pipeline that used to compete with Mercor for senior attention is gone. The senior partner has no protégé to mentor on Tuesday afternoon. So when Mercor's recruiter calls offering $250/hour for "a few evening sessions," the answer is yes.

Stack those three together and you get the picture. Expert knowledge is being extracted on three fronts simultaneously, and the firm's internal succession plan — the one thing that used to keep that knowledge proprietary — has been defunded.

What's Actually Left as a Moat

Not your credentials. The LLM has read every credentialing exam.

Not your published work. Already trained on.

Not your firm's brand. Brands without knowledge transfer become hollow within a decade.

What's left is the specific, structured, unpublished version of how you actually solve problems. The decision trees you've never written down. The 200 edge cases you've seen. The framework you use that has no name because you invented it on a Tuesday in 2017 and never bothered to formalize it.

That is the last unextracted asset. And right now, in mid-2026, you are in exactly one of three positions with respect to it:

1. You let the LLMs have it for free by publishing freely, posting on LinkedIn, answering questions in public forums. Their terms.
2. You sell it hourly to Mercor or a competitor platform. Their terms. You get paid once. They get the asset forever.
3. You package it yourself, own the distribution, own the revenue, and decide who learns it and at what price. Your terms.

There is no fourth option. Neutrality is option one with extra steps.

The Build: What Expert-Owned Knowledge Transfer Looks Like

I built Skill Refinery as the proof-of-category for option three. It's a Knowledge Delivery System — the infrastructure an expert uses to extract their own tacit knowledge, structure it into transferable assets, and deliver it through channels they own.

The stack has four layers.

Capture. Voice-first interview systems that pull the unpublished framework out of the expert's head. Not a course outline. The actual decision logic. The 200 edge cases. Recorded, transcribed, structured.

Structure. AI agents that take 40 hours of raw expert audio and reorganize it into a taxonomy — modules, prerequisites, scenarios, assessments. The work that used to take a curriculum designer six months takes eleven days.

Distribution. Owned channels. Email list. Direct enrollment. No platform middleman taking 50% and learning your students' behavior to build a competitor.

Compound. Each cohort generates new edge cases, new questions, new refinements. The IP gets sharper. The asset appreciates. This is the opposite of the Mercor model, where you deposit your knowledge once into someone else's vault and watch it depreciate to zero in your own account.

The numbers I'm seeing with experts who run this play: $180K-$600K in year-one cohort revenue from knowledge they were giving away in conference talks for free. Time to first revenue: 90-120 days from capture to first paid cohort. Margin: 70-85%, because there's no platform tax.

More importantly — and this is the part nobody puts on a sales page — the expert ends up with a structured, documented version of their own thinking. The same artifact that used to take a 15-year apprenticeship to transfer now exists as a deliverable. You can sell it. You can hand it to a successor. You can use it to train the AI agents inside your own firm instead of feeding Anthropic's.

The Quiet Conviction

Knowledge is a gift before it's a product. The people who taught me what I know didn't bill me for it. They poured it out because that's what stewards do — they pass the thing forward so the next generation can build on it instead of starting from zero.

What I object to is not the transfer. It's the theft of the terms of the transfer. When an expert decides freely to give knowledge away, that's stewardship. When a platform extracts it under the cover of an hourly contract and resells it forever, that's something else.

Own the terms. Then be as generous as you want.

What To Do This Quarter

If you're an expert reading this, three moves matter in the next ninety days.

One. Audit what you're currently giving away for free. Conference talks. LinkedIn posts. Podcast appearances. Mercor sessions. Add up the hours. That's your annual donation to someone else's moat.

Two. Pick the one framework you use that has no name and isn't written down anywhere. That's your asset. Start there.

Three. Decide which of the three economies you're in. Not aspirationally. Actually. Where did your expertise go last month?

The junior hiring collapse is not a hiring story. It's a knowledge transfer story. The pipeline that used to keep expert IP inside firms is gone, and the vacuum is being filled by extraction platforms that pay you once for an asset that compounds for them forever.

The last moat is the structured, owned, distributed version of what's in your head. Build it now or watch it get scraped, hour by hour, by people who understood the math before you did.

I'm writing a book about this — On Whose Terms: The New Expert Economy and the Fight for What You Know. If the thesis resonates, join the launch list.

The systems, the playbooks, and the full Knowledge Delivery System live at mattcretzman.com.

Keep Building,

— Matt

← Back to all posts