video youtu.be Watch ↗

The $200K AI Job That Didn't Exist Last Year

TL;DR

A beginner-facing career pitch (Nate Herk / "AI Automation") that argues the next AI opportunity is not becoming the best builder but becoming your company's in-house AI consultant — the person who decides what to automate and drives its adoption, a de-facto chief AI officer role that mostly has no formal title yet. The setup is the graph's existing thesis reached from the labor-market end: AI has commoditized building ("one person with AI does the work of 3–5"; the AI-agency market that profited on the knowing-problem/knowing-solution gap is now displaced as building goes self-service), so value moves off execution onto judgment — "knowing what to build," the doctor who diagnoses vs the pharmacist who dispenses. The one genuinely operational contribution is a four-step roadmap to create the role from inside: (1) audit your own job and pick tasks that eat real recurring hours and fail safely; (2) automate them and write down the hours saved as proof; (3) make the wins visible in business language and document your prompts/workflows so the team relies on you; (4) graduate from automating annoyances to attacking the business constraint ("if we doubled customers tomorrow, what breaks first?"), then add up the saved hours into one dollar figure and propose the actual role. The market and survey numbers (a $130B automation market; an IBM survey showing chief-AI-officer prevalence rising 26%→76%; a 56% AI-skills pay premium) are the source's claims, flagged for later grounding, not adjudicated here.

Concepts introduced

In-House AI Diagnostician — new. The video's central named thing: the role whose value is deciding what to point AI at plus driving adoption (the "doctor," not the builder "pharmacist"), often with no formal title yet, created from the inside via proof of value. The graph held the judgment thesis (Imagination Constraint) but had no page for it as an organizational role.
Annoyances vs Constraints — new. Theory-of-constraints for AI adoption: automating annoyances makes you helpful; attacking the binding business constraint is what makes real money. The graph had neither the annoyance/constraint distinction nor the "if we doubled volume, what breaks first?" probe.
Low-Blast-Radius First — new. The first-automation selection heuristic: pick tasks that eat real recurring hours and fail safely (human in the loop), not the most annoying one. A concrete task-selection rule the graph didn't hold.
Imagination Constraint — corroborated. "Knowing what to build > knowing how to build; value is in judgment/taste/deciding what to point AI at" independently reconverges from the career altitude.
Execution Commoditization — corroborated. "One person does the work of 3–5," "cost/value of development is dropping," and an AI-agency service commoditizing itself are this concept seen in headcount and in a service business.
Capability Overhang — corroborated (adoption half). "Adoption is another huge problem — change management, stakeholder communication" is the human face of the payoff lagging until the org absorbs the capability.

Held, not dropped (touched, but not warranting their own page yet):

AI-native / early-mover timing — "get in early; most won't realize the job exists until the market's saturated." Already held graph-wide as ai-native-adoption-window (see How Claude Is Creating a New Generation of Millionaires); not re-spun.
Speak in business outcomes, not tools — "don't say 'I used ChatGPT'; say 'this saved us 8 hours before the quarterly report.'" A communication best-practice; folded into In-House AI Diagnostician as a claim rather than its own page. Hold as business-framed-wins if a second source treats it standalone.
Sanctioned-automation boundary (regulated industries) — "don't throw AI at sensitive data; don't automate without permission; practice on dummy data." A governance caveat; hold as sanctioned-automation-boundary, spin out on a second source.
Personal-trainer / "get your own body in shape first" analogy — rhetorical support for prove-it-on- yourself-first; not a concept.
Free-community / course CTA — promotional, dropped.

Key claims

When building commoditizes, the value and pay inside a company move to whoever decides what to build and drives its adoption — the diagnostician, not the builder. principle → In-House AI Diagnostician, Imagination Constraint. Durable: the scarce complement to cheap execution is problem-selection.
The builder is the pharmacist (dispenses what you asked for); the in-house AI consultant is the doctor (figures out what you need) — and the doctor gets paid the real money. observation → In-House AI Diagnostician. The source's framing metaphor for where value sits.
Automating annoyances makes you helpful; removing the constraint the whole business is stuck behind is how you make the company real money. principle → Annoyances vs Constraints.
Find the constraint by asking "if we doubled our customers tomorrow, what process would break first?" rather than "what's annoying / eats a few hours?" best practice → Annoyances vs Constraints. Context: an org-scale audit past the learning phase.
Pick first automations by two filters at once: high recurring time-cost and low blast radius (safe to be wrong, human in the loop); the most annoying task isn't necessarily the right first one. best practice → Low-Blast-Radius First. Context: early automation with a human reviewer; inverts for high-stakes/irreversible tasks.
Being able to build is necessary but the smaller piece; knowing what to build is larger — yet "if you can't actually build the solutions, nothing else matters." best practice → In-House AI Diagnostician. Judgment and delivery are a product, not a substitute pair.
Prove value on your own work first, quantify saved hours into one business figure ("these five automations = a full-time hire per year"), then propose the role — you create it from inside, you don't find it on LinkedIn. best practice → In-House AI Diagnostician. Context: an employee with no posted AI position.
Adoption/change-management is a first-class part of the role — a built automation nobody uses delivers no value. principle → Capability Overhang, In-House AI Diagnostician.

Check-worthy source claims (attributed, groundable, not adjudicated — a later grounding pass can verify):

~76% of organizations in an IBM survey of ~2,000 CEOs now have some kind of chief AI officer, up from ~26% a year earlier. observation — on-screen IBM article ("CEOs are Reshaping C-suite Roles for the AI Era," May 2026) + narration.
People with AI skills are paid ~56% more than a colleague doing the same job without them. observation — narration.
The AI-automation-services market grew to roughly $130 billion. observation — narration.
Chegg's stock crashed almost 50% in a single day (2023) and it attributed the decline to ChatGPT. observation — narration over the Chegg homepage.
Sam Altman said "the idea guys are about to have their day in the sun"; OpenAI and Anthropic have said their models are getting "scary good." observation — narration, attributed quotes.

Why this corroborates the graph (and what's new)

The dominant stance is corroborates: a third Nate Herk source independently reaches the graph's established thesis — value has left execution (Execution Commoditization) and landed on deciding what to build (Imagination Constraint) — this time from the career-positioning altitude rather than the strategy-essay or millionaire-hype altitudes the graph already holds (see You Can't Compete on Cheap Models Anymore, How Claude Is Creating a New Generation of Millionaires). That the same conclusion survives translation into "here's the job to aim for" is itself the signal worth recording; the concepts are not re-created, only backlinked.

Three secondary novel stances carry the new material: In-House AI Diagnostician (the thesis crystallized into an organizational role, with adoption added — the graph's first page on the human who captures the displaced value), Annoyances vs Constraints (theory-of-constraints applied to what to automate, with the "double the volume" diagnostic), and Low-Blast-Radius First (a concrete first-automation selection heuristic). A builds_on note for Capability Overhang: the video adds the human-labor face of the overhang — adoption and change-management as first-class work — to a concept that had described the org-redesign requirement structurally but not who drives it.

Illustrated walkthrough

This is a talking-head explainer over stock b-roll and motion-graphic cards; the beats below track the on-screen argument at its natural altitude.