The AI Skill Nobody's Teaching (That Will Matter Most In 2026)

It's not prompting. It's not automation. It's something far more human.

New here? I'm James — I help founders and creators build AI systems that make them impossible to replace. I teach you to build infrastructure around your thinking, your voice, and your specific bottlenecks.

This piece (co-written with Mia Kiraki) digs into the skill that'll separate the commoditized from the dangerous in 2026: AI translation — knowing where AI fits and where it doesn't.

If you want the deeper foundation, start with Unpromptability: 5 Steps to Become Irreplaceable in the Age of AI.

Unfortunately, 99% of people will remain “AI users.”

Not that there’s anything wrong with that, but most of the people here want more than just to use AI. They want to solve problems, build businesses, do things that matter with this powerful tool. They want to be amplified.

Nevertheless, if you don’t understand one fundamental problem and build the skill to combat it, you’re going to get stuck as an AI user. Same as everyone else.

This will leave you fragile as you start the new year.

Why you need to stop being just an “AI User”

This year saw significant advancements in AI and its applications -- but it’s just the warm-up. In the next months, we’re poised to enter an AI era that’s never been seen before.

But only a fraction of the population will be able to truly take advantage of it.

That’s because as AI becomes more accessible, every industry now has access to the same tools. Skill with AI isn’t the only advantage anymore. Interpretation will also become crucial.

Think about it. Two creators can use the same LLM. One produces forgettable content at scale. The other builds systems that sharpen their thinking, tap into audience needs, and convert them into customers -- all while protecting their unique voice. Same tool, wildly different outcomes.

The gap isn’t access. It’s translation.

The winners of 2026 are those who turn complexity into clarity—who look at a messy human problem and know precisely where AI fits (and where it doesn’t).

This is why I invited Mia Kiraki into this edition. She’s been thinking deeply about this translation problem, and what she’s written is worth your attention.

Here are her insights.

AI translation: The massive gap nobody’s talking about

We all know there is absolutely no AI shortage out there. There are tens of LLMs. Hundreds of ways to combine them all. Thousands of capabilities, thousands of workflows.

What we have is a translation crisis.

Walk into any company right now and you’ll find a pattern. There are people who’ve heard AI can transform everything and a massive gap between the hype and anything that actually works.

Everyone knows the tools exist but almost no one knows how to diagnose which business problem the tools should solve or how to write them into their workflow without killing what makes their work valuable.

I’m here to tell you that what you need is the skills of an AI translator: to be a person who can look at a messy, human business bottleneck and build an AI solution that makes it faster while protecting what makes the human work valuable.

Even though you’ll be using AI, you won’t hand over your job. You’ll know exactly where AI amplifies your creative process while your intellectual signature stays intact.

The three AI translation problems that need solving

Let me show you what real AI translation looks like by walking through the three challenges that separate translators from amateurs.

Diagnosing the real bottlenecks

A lot of people misdiagnose their problems. They think the bottleneck is “writing takes too long” so they ask AI to write for them. Then they spend three hours editing the slop back into something usable.

That’s just trading one problem for another.

When you put on the hat of a translator, you audit your workflow to find the highest-friction decision point.

Here’s an example from my own process. I went through a phase when I thought my bottleneck was producing copy and messaging materials. That was wrong. My real bottleneck was deciding which customer to target. I spent days wrestling with headlines and second-guessing concepts, all because I hadn’t clarified my audience first.

So back then, I’ve built a prompt that took my messy braind dump and forced me to answer 4 hard questions (this is more broad than just putting together messaging for an audience, so it can apply to more use cases):

• What’s the non-obvious insight here?

• What cultural references makes this memorable?

• WHat is the counterargument I’m afraid to address?

• Who else is saying something similar and how is my angle different?

The AI would cross-examine my ideas until I find the one that is differentiated.

Here’s the prompt:

You are a strategic interrogator. I’m going to give you a rough idea  for [project type]. Your job is to stress-test it by asking me the questions I’m avoiding.

For each concept I present, challenge me on:

1. Differentiation: “Who else is saying this, and why would someone choose YOUR version?”

2. Depth: “What’s the counterintuitive insight here that makes this worth someone’s time?”

3. Proof: “What evidence or example proves this actually works?”

4. Memorability: “What’s the cultural hook or metaphor that makes this stick?”

If the idea is generic, tell me. If I’m hiding from the hard question, call it out.

Here’s my rough idea: [INSERT YOUR CONCEPT]

Keeping your voice while scaling output

The second failure mode is: people use AI to produce more content but it all sounds like everyone else. And even worse, it’s obvious AI slop for the trained eyes.

AI makes it easy to be productive and nearly impossible to be distinct. I love to say this: the default mode is the average. If you don’t actively fight it, you’ll regress toward mediocrity (at scale, true, so pick your poison!)

As a translator, you need to systematize your taste before you scale.

I never run my work through one clever prompt and call it a system. For most of what I ship, there’s a 30+ step pipeline behind the scenes (layered checks, filters, etc). This is for letting my creative judgement always stay in control.

Most of those steps are creative constraints and at the core of that pipeline sits an operating system for my voice. Every AI-assisted work of mine passes through it before it gets anywhere near “publish”.

The full version has dozens of rules and sub-routines, but the fundamental looks like this:

• Define non-negotiables;

• Define signature moves;

• Formalize them into checks;

• Add consequences.

Since we’re speaking mostly about content in this post, here’s a simplified version of the content editor layer you can copy, stripped down to the basics (most of the in-depth layer checks are proprietary and they should. That’s the point: you, as a translator, don’t just use AI but build a private infrastructure of taste).

With that in mind, make sure you use it as inspiration and build your own workflow:

Your job is not to be nice. Your job is to enforce my editorial standards so nothing generic or beige makes it through.

I’m going to give you a draft. Run it through these checks:

INSTANT FAILURES (must be rewritten or deleted):

- Metaphors about journeys, roadmaps, or “the landscape”

- Sentences starting with “In today’s world” or “As we all know”

- Corporate jargon (synergies, leverage, circle back, deep dive, ecosystem)

- Obvious transitions (Furthermore, Additionally, Moreover, In conclusion)

- Advice so generic it could apply to any industry or person

REQUIRED SIGNALS (must appear consistently):

- At least one specific, concrete example per major section

- A clear point of view (no neutral, “on the one hand / on the other” tone)

- Varied sentence length (short, punchy lines mixed with longer ones)

For each issue you find:

1. Quote the original sentence or paragraph.

2. Explain in one line why it fails the standard.

3. Propose a rewrite that matches the voice and raises the bar.

If an entire section feels beige or interchangeable, say:

“THIS SECTION IS GENERIC. RECONSIDER THE CORE IDEA.”

Now, here’s the draft:

[INSERT YOUR CONTENT]

Build systems that get smarter

AI is simple: input question, get answer, repeat.

But you won’t ever learn anything that way. You’re outsourcing the work, which means you’re also outsourcing the skill development.

As a translator, you’ll build workflows that challenge your thinking instead of replacing it.

Example: I almost never use AI to generate content ideas. I use it to expose my blind spots and help put the ins and outs of my brain and experiences on paper.

I write a ton about culture references, characters I love and the like. And one of my favourite characters is Lord Varys from Game of Thrones.

Lord Varys knew: the story always moves faster than the sword

I use him a lot to come up with creative prompt ideas, and one of my favourite Varys prompts is “The Varys System”. I feed it my content strategy and it role-plays as a competitor analyst, showing me gaps in my positioning and the obvious signals I’m ignoring.

Here’s how it works:

You are Varys from Game of Thrones, a master of intelligence and strategic positioning. I’m going to share my content strategy, and your job is to analyze it like a competitor would.

My strategy: [PASTE YOUR CONTENT PLAN, POSITIONING, OR RECENT WORK]

Now, give me a strategic intelligence report:

1. BLIND SPOTS: What obvious angles am I completely missing?

2. COMPETITIVE GAPS: Where are my competitors zigging while I’m zagging (and is that good or bad)?

3. MARKET SIGNALS: What trends or pain points am I ignoring that I should address?

4. POSITIONING RISKS: Where does my differentiation feel forced or unsustainable?

5. STRATEGIC MOVES: What’s one counterintuitive shift I should consider?

Don’t be gentle. Your allegiance is to the truth, not my ego.

The AI forces me here to think harder about my strategy and every session sharpens my judgment.

Every time I run this workflow, I learn something about my own biases. I get faster, but I also get sharper.

The bottom line: by mid-2026, “I use AI” will mean nothing. The market will pay for people who can translate business chaos into AI clarity without killing what makes their work valuable.

The market’s about to split into two groups: people who got commoditized by their tools, and people who got dangerous with them. Pick which side sounds more fun and I hope to see you on the good side!

The skill that will pay dividends in 2026

Tools won’t differentiate anyone next year. Everyone will have them. Everyone will know the basics.

What separates the commoditized from the dangerous is this: the ability to see where AI fits and where it doesn’t.

The discernment, born from deep thinking and irreplaceable personal experience, is to diagnose real bottlenecks instead of surface symptoms. To build systems that make you sharper and smarter, not just faster.

That’s translation. That’s the skill.

Mia nailed it: clarity is the new leverage. Not more prompts. Not more tools. The ability to look at chaos and know exactly what to do with it.

So here’s the question worth sitting with: where are YOUR bottlenecks? Not the ones you assume. The real ones—the decision points where you lose hours, where your thinking gets fuzzy, where you reach for AI as a crutch instead of a lever.

Find those. Build systems around them. Protect what makes your work yours.

That’s how you stop simply being an AI user.

That’s how you become a translator.