Why Modern Creator?

Nate Herk | AI Automation · YouTube

How to Become Your Company's Go-To AI Person

Why the bigger opportunity in AI right now isn't starting an agency, but becoming the person inside a business who turns AI budget into results.

Posted

August 18th

yesterday

Duration

08:04

Format

Essay

educational

Views

15.6K

625 likes

Big Idea

The argument in one line.

Because 95% of enterprise AI pilots fail to show a return, the fastest-growing, best-paid role in business right now is the in-house 'AI person' who can turn AI spending into measurable results.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You work inside a company rather than run an agency, and want to become the go-to person for AI-built fixes instead of just talking about AI in meetings.
You're comfortable using a tool like Claude to build small automations (reports, inbox sorting, data cleanup) and want a repeatable process for proving their value.
You're weighing an AI agency against an in-house AI role and want the pay and career case for the second path.

SKIP IF…

You want a step-by-step build tutorial for a specific automation. This is a positioning and career strategy video, not a build walkthrough.
You've already built a scaled, revenue-generating AI agency. The video is aimed at people establishing themselves for the first time.

TL;DR

The full version, fast.

Most people chasing AI income are told to start an agency, but a bigger opportunity is becoming a company's in-house 'AI person' who actually builds automations instead of just prompting ChatGPT. AI-skilled workers already earn a 62% pay premium, and roles like forward deployed engineer and chief AI officer are exploding in both demand and pay. The catch: a 2025 MIT study found 95% of enterprise AI pilots show no measurable return, so companies desperately need someone who can convert AI spend into results. The playbook is three phases: position yourself as a builder inside one team, prove your value by naming a number before you build a fix and recording the before-and-after, then become impossible to replace by pinning wins to your name and attacking the business's real growth bottleneck.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →

Chapters

Where the time goes.

00:00 – 00:33

01 · The agency trap

Opens with the standard AI-agency pitch, then pivots: the bigger opportunity is becoming a company's in-house AI person. Splits AI users at work into two types (question-askers vs. builders) and previews the roadmap.

00:33 – 01:14

02 · Two types

Defines the AI person as the builder type, who finds problems worth automating and builds the fix, and can do it either as an outside service provider or as an in-house hire.

01:14 – 02:16

03 · The money?

Cites PwC's 2026 report: a 62% pay premium for AI skills (up from 25%), forward deployed engineer postings up from ~640 to 5,000+ a year at a ~$210K median (Palantir, OpenAI, Anthropic hiring), chief AI officer at a $1.6M median with postings up 478%, and AI job titles tripling since 2022.

02:16 – 03:31

04 · What's missing?

Names the gap: companies have budget and pressure to use AI, but a July 2025 MIT study found 95% of enterprise AI pilots show no measurable return, and McKinsey found only 7% of companies have scaled AI despite 88% using it somewhere. Plugs his free Skool community's AI-pricing guide.

03:31 – 04:29

05 · Start here

Phase 1, position yourself: be the visible builder on one team, pick a narrow niche instead of the whole company, and market by helping coworkers and asking your boss what eats the most time.

04:29 – 06:15

06 · Which number?

Phase 2, prove your value: pick one cheap-to-get-wrong, repeatable task, only use company-approved tools, name the number you're moving (time saved, mistakes cut, money made) before building, then build with real docs, record a before-and-after, and deliver measured proof.

06:15 – 08:04

07 · Real bottleneck

Phase 3, become impossible to replace: pin every win to your name, then raise the altitude by attacking the business's real constraint (supply vs. demand constrained), clearing bottleneck after bottleneck until the role can't be cut.

Atomic Insights

Lines worth screenshotting.

AI-skilled workers now earn a 62% wage premium over peers doing the same job without AI skills, up from just 25% two years ago.
Forward deployed engineer job postings jumped from about 640 to over 5,000 in a single year, with a median salary near $210,000 at Palantir.
Chief AI officer, a title that barely existed three years ago, now carries a median pay of $1.6 million at companies that disclose it.
A July 2025 MIT study found 95% of enterprise AI pilots delivered no measurable financial return.
88% of companies say they use AI somewhere, but only 7% have actually scaled it across the business.
The number of job titles mentioning AI has tripled since 2022, and almost two-thirds of those roles sit outside tech.
The gap between a builder and a consultant is a single habit: naming the exact number you're trying to move before you start building.
Every workplace automation win falls into one of three buckets: time saved, mistakes cut, or money made.
A quick test for whether a business is supply or demand constrained: ask what breaks if the company doubled its customers tomorrow.
A win that saves a company real money doesn't automatically get credited to the person who built it unless that person actively connects it back to their name.

Takeaway

Three phases that make you the AI person a business can't replace.

AI CAREER ROADMAP

Becoming a company's in-house AI person beats starting an agency because the pay premium is exploding while almost no company has proven it can turn AI budget into results.

01The agency trap

The 'start an AI agency and pitch cold-outreach automations' playbook is getting crowded; becoming the go-to AI person inside a business is a less-saturated path to the same money.
There are two kinds of AI users at work: people who ask a question and stop, and people who build something (an agent, a workflow, a data-cleanup system). The second type is who this roadmap targets.

02Two types

You can play the AI person role two ways: as an outside service provider going business to business, or as an in-house hire embedded in one company.
Choose deliberately to build things other people in the business actually use, not just tools you experiment with for yourself.

03The money?

Workers with AI skills earned a 62% pay premium in PwC's 2026 report, up from 25% two years earlier, so the premium is accelerating, not leveling off.
Forward deployed engineer postings jumped from about 640 to over 5,000 in a year, with a roughly $210,000 median salary at Palantir, and OpenAI and Anthropic are hiring for the same role.
Chief AI officer, a title that barely existed three years ago, now carries a median pay of $1.6 million at companies that disclose it, with postings up 478% in one year.
AI-related job titles have tripled since 2022, and nearly two-thirds of them sit outside tech, in healthcare, marketing, logistics, and management.

04What's missing?

The gap isn't budget or willingness, most companies already have both. A 2025 MIT study found 95% of enterprise AI pilots delivered no measurable financial return.
McKinsey found a similar pattern: 88% of companies say they use AI somewhere, but only 7% have actually scaled it across the business.
That gap is the opening: any company spending on AI without results needs someone who can turn that spend into an actual outcome.

05Start here

Phase one is positioning: be the person who visibly builds things with AI, not the person who just talks about AI in meetings.
Pick one team and a few workflows to start. Don't try to be the AI person for the whole company on day one.
Make your work visible by helping coworkers with their most annoying task and asking your boss what eats the most time in the department, so you build what people actually asked for.

06Which number?

Start with a task that's expensive and repeatable and where a mistake doesn't hurt anyone: weekly reports, meeting notes, inbox sorting, data cleanup.
Only use AI tools your company has approved, and never put company or customer data into anything unapproved.
Before building anything, name the exact number you're trying to move: time saved, mistakes cut, or money made. That's what separates a builder from a consultant.
Build the fix with the real documents behind the task, record a before-and-after, and deliver proof, not just a promise.

07Real bottleneck

A stack of wins is useless if nobody connects them to you. Frame every win as the business's win, but make clear it happened because of the system you built.
Saving time is useful but doesn't grow a business the way attacking its biggest constraint does. Every company only grows as fast as its worst bottleneck allows.
Ask what breaks if the company doubled its customers tomorrow. If everything falls apart, the business is supply constrained; if it could handle it fine, it's demand constrained.
Clearing one bottleneck reveals the next one. The AI person who keeps attacking constraints becomes the person a business can't afford to lose.

Glossary

Terms worth knowing.

Forward deployed engineer: A technical role, often embedded directly with enterprise clients, that builds and implements AI or software solutions inside a customer's business rather than shipping a generic product.
AI pilot: A small-scale trial of an AI tool or workflow a company runs to test results before deciding whether to roll it out company-wide.
Supply constrained: A business that can't produce or deliver enough to keep up with existing customer demand.
Demand constrained: A business that could easily handle far more customers than it currently has, so growth is capped by how many customers come in, not by delivery capacity.

Resources

Things they pointed at.

01:18linkPwC 2026 workforce report (AI skills wage premium)

02:30linkMIT July 2025 study on enterprise AI pilots

02:35linkMcKinsey enterprise AI adoption study

03:10linkAI Automation Society free Skool community + AI-pricing guide ↗

Quotables

Lines you could clip.

00:22

“The real opportunity for most people right now is to become the AI person.”

clean thesis statement, works as a cold open→ TikTok hook↗ Tweet quote

02:35

“They found that 95% of company AI pilots delivered no measurable return at all.”

shocking, sourced stat that reframes the whole video→ IG reel cold open↗ Tweet quote

05:00

“A builder is just gonna build. But a consultant picks one number and actually moves it.”

sharp distinction, quotable framework line→ newsletter pull-quote↗ Tweet quote

07:45

“You're not an employee who's good with AI. You're the reason the business is growing.”

strong closing punch line, identity-shift framing→ TikTok hook↗ Tweet quote

The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

metaphor

Everybody's telling you the same way to make money with AI in 2026, which is to start an AI agency, find clients, and then sell them AI automations. But the opportunity to build a profitable AI agency is shifting. The real opportunity for most people right now is to become the AI person.

And right now, this is the closest thing to a job that can't be replaced. I actually used to be an AI person at one of the biggest banks in the world just a few years ago. Now that position has seen a huge increase in demand over the last couple of years and it will keep growing.

So in this video, I'll break down what the AI person actually is, why every business is going to be desperate for one, and the exact road map for you to become one. So let's begin. Alright.

So what actually is the AI person? Right now, there's two types of people using AI at work. The first type opens up ChatGPT or Copilot, asks a couple questions, gets an answer, and that's pretty much where it ends.

While the second type knows how to take AI and actually build something with it, whether that's an agent that runs the support inbox or a workflow that writes their weekly report or a system that cleans up the data before anyone touches it. The AI person is the second type of person. You're the go to inside a business for anything AI.

The one who finds the problems worth automating and then builds the fix. And you can do this one of two ways. You can do it as your own thing, going business to business and selling it as a service, or in house where you become the AI person inside of a company.

And that seed is worth going after because it's the fastest growing, best paid ground in the entire job market right now. I mean that with actual numbers. PwC tracks over a billion job ads every single year.

And in their 2026 report, workers with AI skills are getting paid a 62% premium over people doing the exact same job without those AI skills. And a couple of years ago, that premium was only 25%, but it more than doubled and it's still going up. The roles at the very top pay a lot more than that.

The forward deployed engineer, which is basically the AI person who's heavily focused on building, went from around 640 job postings to over 5,000 in a single year. And Palantir pays them about $210 at the median, and OpenAI and Anthropic are also hiring the same role right now.

And chief AI officer, a title that barely existed three years ago, is paying a median of 1,600,000 at the companies that are actually disclosing this kind of stuff. Those postings were up 478% in one year.

The number of job titles that even mentioned AI have tripled since 2022, and almost two thirds of them are outside of tech. So health care, marketing, logistics, management.

So being the AI person isn't just some niche tech job. It's turning into a requirement in every company and every department, and the earlier that you can become the AI person, the faster you can move up. So that's the value to you.

But the reason that every business is about to need this person comes down to one single gap. Almost every company already knows that AI matters. You know, they've got the budget, and they're under real pressure to use it so that they don't fall behind their competitors.

But the problem is almost none of them have pulled it off successfully. I'm sure you guys have all heard that MIT study. They ran a study on enterprise AI products in 2025, and they found that 95% of company AI pilots delivered no measurable return at all.

And then Mackenzie found basically the same thing. 88% of companies say that they're using AI somewhere, but only 7% of companies have actually scaled it across the business. So what we're seeing is the budget's there, the pressure's there, the results are not there.

Every one of these companies needs someone who can walk in and turn the AI spend into an actual result. And right now, that seat is pretty empty. And real quick, before we jump into the actual road map, I wanna let you know that have a full guide on how to price AI workflows in my free school community, which you can access for completely free using the link in the description.

I also published a video alongside that guide where I go over all these different scenarios and methods, which I will tag right up here if you wanna check that out. Now whether you wanna sell your services to businesses or you wanna be the in house AI person, understanding what I talk about in that video is really important because it's all about proving the value that these systems will create and justifying the expense, which is how if you're the in house AI person, you're able to go ask for, you know, bigger budgets and more resources for your projects that you wanna take on.

So anyways, let's get back to the video. Here is the actual road map to becoming the in house AI person, and I broke it into three phases. So phase one is just to position yourself.

You start as the builder, and that just means you're the one actually making things with AI, not the person just talking about AI in the meetings. Then you're gonna go ahead and pick your niche. So don't try to be the AI person for the whole company on day one.

Just start with one team with a few workflows. Choose the team you're already on and make that your patch of ground. Then you start to market.

Right? Inside a company, that just means you make your work visible. So you help coworkers with their most annoying task.

You ask your boss what eats up the most time in the department. And now you're not guessing what to build. You're having people on the team actually tell you what would be valuable to them.

Then we move on to phase two, which is proving your value. So you start with one task off that list, you know, the most expensive, the most repeatable, where if AI gets a little bit wrong, nobody gets hurt. So these could be weekly reports, meeting notes, sorting the inbox, cleaning up data, just the stuff that's boring but really repetitive.

And one quick rule before you touch anything, only use the AI tools that your company actually allows, and never put company or customer data into something that isn't approved. Here's the most important part. Before you build, you have to name the number that you're aiming to move.

This is the whole difference between a builder and a consultant because a builder is just gonna, you know, build. They're just gonna make some things that look cool. But a consultant picks one number and actually moves it, and that number almost always falls into one of three buckets, which are time saved, mistakes cut, or money made.

So for example, a Friday report, that number that you're trying to move is maybe four hours at each every week and turning that into something like twenty minutes. But these numbers can also get more specific, and honestly, the more specific the better, like organic form submissions per week or average response time or refund percentage.

And then once you have the number, you build a fix to move that number and then you record it. You give Claude the real docs behind the task, the last few reports, the template, and you work with it until it nails the report every time. And then you record a quick before and after video or case study.

So, hey, you know, this used to take four hours, but watch me do this in ten minutes because I built this AI system. And then you deliver it for real and you prove that the number moved. Not just saying, hey, know, this should save you some time.

You actually measure it. Four hours down to twenty minutes is three and a half hours back every single week, and now you've got proof, not just a promise. And then you basically just run that same loop on the next task and the one after that until you've built up a whole stack of wins with real numbers attached, and then you move into phase three, which is to become impossible to replace.

Because if you have a big stack of wins, but they're you know, you're not doing anything with them, then it's useless. Because you could build a bunch of systems, they're actually helping the business, but if nobody knows, then what's the point? So phase two is about getting the results.

Phase three is about making sure those results actually get pinned to your name. Because here's what usually happens. Like I said, the business starts saving real money and everybody just assumes someone upstairs will notice and connect it back to you, but usually they won't.

And it just kind of disappears into, oh, yeah. We had a great quarter. So what you have to do is connect it.

Every win that you bring to the team, you frame it as the business's win, but you make it undeniable that it came because of the systems that you built. And when you're able to show them all of this proof, all of these outcomes, that's the math that gets a role created for you, and then you raise the altitude.

Because saving people time is useful, but it's not ultimately like what really, really grows a business. What really grows a business is attacking constraints. A company only grows as fast as its single biggest bottleneck lets it.

And at the highest level, every business is one of two things. It's either supply constrained where they can't produce or deliver enough to keep up with demand, or it's demand constrained where they could handle way more customers. They just don't have enough coming in.

And a quick way to tell which one you're in is just to ask what breaks if the company doubled its customers tomorrow. If everything falls over, then you're supply constrained. And if you can handle them just fine and you're just not there, then you're demand constrained.

And this is truly where you stop being just like, hey, the AI guy, the AI automation guy, and you become the person with the perspective. You walk into the room and you say, this is our real bottleneck. This is the exact thing capping our growth, and here's the AI system that I'd like to build to attack this constraint.

Then you build it, you prove that that constraint actually opened up, same loop as phase two, a much bigger number at a much larger scale. And the second that bottleneck clears, guess what happens? A new one takes its place.

So then you go attack that one too. And you just keep doing that constraint after constraint, and that's the person that a business physically doesn't wanna have to replace. You're not shaving a few hours off someone's week anymore.

You're the one moving the thing that their entire growth is stuck behind over and over. And at that point, they don't really have a choice. The budget gets bigger for your projects.

They put more people under you so you can move faster, and the resources just start showing up. You're not an employee who's good with AI. You're the reason the business is growing.

So that's the full road map. Position yourself, prove your value, and become impossible to replace. But there is one catch.

You can follow every step in this road map, but if you can't actually build the solutions with AI, then none of this works. The good news is you can learn how to build all of it for free in my community. Full courses, all the tools, and every resource that I use in my videos.

I also put together a resource free guide today that covers everything that I just talked about in this video, and it's in my free school community as well. And if you get stuck, you've got me and the whole community in there to help you out. So, anyways, that's gonna do it for this one.

And if you got something out of it, please give it a like. It helps me out a ton. And as always, I appreciate you guys making it to the end of the video.

I'll see you on the next one. Thanks, everyone.

The Hook

The bait, then the rug-pull.

Everyone's chasing the same AI agency playbook right now, but Nate Herk argues the real money is in becoming the person inside a business who can actually turn AI spend into a result, a role he calls 'the AI person.'

Frameworks

Named ideas worth stealing.

03:31list

The 3-Phase AI Person Roadmap

Position yourself
Prove your value
Become impossible to replace

The full career arc from unknown builder to indispensable in-house AI resource: get visible on one team, attach hard numbers to your fixes, then attack the company's actual growth bottleneck.

Steal forany internal AI-adoption pitch or a freelance AI consultant's positioning

05:00list

Name the Number

Time saved
Mistakes cut
Money made

Before building anything, pick the one metric the fix is supposed to move. This is the line between a builder (makes things that look cool) and a consultant (moves a number and proves it).

Steal forframing any automation pitch to a boss or client before starting the build

06:30concept

Supply vs. Demand Constrained

Every business is capped by one of two things: it can't deliver fast enough (supply constrained) or it can't get enough customers in the door (demand constrained). Ask what breaks if customers doubled overnight to tell which one you're in.

Steal fordiagnosing where an AI system would create the most leverage inside any business

CTA Breakdown

How they asked for the click.

VERBAL ASK

07:40link

“The good news is you can learn how to build all of it for free in my community. Full courses, all the tools, and every resource that I use in my videos.”

Soft-pitches the free Skool community twice, once mid-video right before the roadmap and once in the outro, framed as a resource rather than a hard sell

MENTIONED ON CAMERA

03:10linkAI Automation Society free Skool community + AI-pricing guide ↗

FROM THE DESCRIPTION

PRIMARY CTAWhere the creator wants you to go next.

AFFILIATECommission earned if you click.

OTHER LINKSAlso linked in the description.

Storyboard

Visual structure at a glance.

hook / title card

hookhook / title card00:00

AI users vs AI builders

valueAI users vs AI builders00:33

salary data

valuesalary data01:14

95% of pilots fail

value95% of pilots fail02:16

pick a niche

valuepick a niche03:31

name the number

valuename the number04:35

supply vs demand

valuesupply vs demand06:15

roadmap recap / CTA

ctaroadmap recap / CTA07:40

Frame Gallery

Visual moments.

hook / title card

Frame at 00:06 from How to Become Your Company's Go-To AI Person

Frame at 00:13 from How to Become Your Company's Go-To AI Person

Frame at 00:21 from How to Become Your Company's Go-To AI Person

Frame at 00:29 from How to Become Your Company's Go-To AI Person

Frame at 00:33 from How to Become Your Company's Go-To AI Person

Watch next

More from this channel + related breakdowns.

14:33

Nate Herk | AI Automation · Tutorial

Finally. Agent Loops Clearly Explained.

A 14-minute demystification of agent loops for non-hardcore-coders: what they are, why the done-check matters most, and three live demos that prove loops get you closer — not perfect.

June 19th

20:32

Nate Herk | AI Automation · Demo

Grok Bot Is For Real: What You Need to Know

A hands-on tour of a synced, phone-controllable AI agent team — agent computers, teachable skills, scheduled routines, event triggers, and where it stops making sense versus Claude Code or Codex.

August 12th

25:03

Nate Herk | AI Automation · Tutorial

5 Hacks to Instantly Level Up Your AI OS

Nate Herk breaks down the four ways an AI operating system's context quietly goes wrong, then walks through the five habits that keep a growing second brain accurate instead of confidently wrong.

July 23rd

09:36

Nate Herk | AI Automation · Talking Head

How Claude is Creating a New Generation of Millionaires

A single founder makes the case that Claude Code has erased the cost of building software, using a three-person team's state government contract as proof.

July 3rd

28:12

Nate Herk | AI Automation · Tutorial

I asked Claude Code to make me as much money as possible

Four prompt-layer upgrades that fix the documented failure modes quietly killing your Claude output quality.

June 25th

16:53

Nate Herk | AI Automation · Tutorial

So You Learned Claude, Now What?

A 17-minute career roadmap arguing that the next move for anyone who can build with AI is to stop being a builder and start being a consultant — with a four-step playbook to do it without quitting your job.

June 22nd