The most valuable skill in the age of AI is communicating your ideas and your vision clearly. AI tools now give everyone a Wozniak-level builder on call, so building stops being the bottleneck. The people who can say exactly what they want built, and who have the taste to judge the result, get the most out of it.
This post expands a 102-second video I published on September 19, 2026. It's embedded below.
Key takeaways
- Clear communicators win the AI era. People who can express their ideas and vision in a clear, objective way will be the most leveraged group of workers during the AI revolution.
- The meme explains it. "A supreme shape rotator can only rotate shapes, but a supreme wordcel can rotate shape rotators."
- Jobs and Wozniak are the model. Woz could build anything in 1976. He built the right things because someone next to him could say clearly what the product should be.
- Everyone now has a Woz. You can prompt Astra or Fable and it creates things from scratch. What's scarce is the person who knows what to ask for.
Watch it on YouTube · Instagram · LinkedIn · X
A few months ago a friend and I were talking about Steve Jobs and Steve Wozniak, and about a meme that was going around on Twitter. The conversation stuck with me, because it's the cleanest explanation I've heard of who actually benefits from AI tools getting this good.
What is the most valuable skill in the age of AI?
Communicating your ideas and your vision in a clear, objective way. I think the people who can do that are going to be the most leveraged group of workers during the AI revolution.
Here's the mechanism. For most of history, having an idea was cheap and building it was expensive. You needed engineers, designers, editors, a team. AI tools are collapsing the cost of the building part. When building gets cheap, the expensive part moves upstream: knowing what should exist, and describing it precisely enough that a builder (human or AI) can make it.
That second part is a communication skill. It always was. It just used to be hidden behind the cost of the build.
What do "shape rotator" and "wordcel" mean?
A shape rotator is an engineer, a builder, someone who creates things. A wordcel is someone very articulate, someone who can communicate things clearly.
Both are internet slang. According to Know Your Meme, the terms first showed up on 4chan in June 2019 and were popularized on Twitter by the user roon in late 2021. They started as insults: the wordcel is good with words and not much else, the shape rotator is good at math and can't explain himself.
The version I care about is this tweet from David Sun, posted over a photo of Jobs and Wozniak:
"A supreme shape rotator can only rotate shapes, but a supreme wordcel can rotate shape rotators."
Read that twice. The best builder in the world can only build. The best communicator can point builders at the right thing.
Why did Steve Wozniak need Steve Jobs?
Because a great builder still needs someone who knows where the product is going and can say it clearly.
Woz was the shape rotator. He's the one who designed the early Apple computers, and he was a fantastic engineer by any standard. But he was only empowered to create the products he created because he had someone next to him who could express the vision very clearly, who had very good taste, who knew where he was trying to go with the product, and who could communicate that so the engineers could build it.
Jobs was the wordcel in the good sense of the word. He didn't design the circuit board. He made sure the circuit board was for something.
Why does AI make clear communication more valuable?
Because everyone now has access to a Woz, and almost nobody has practiced being the Jobs.
This was my friend's argument, and I think he's right. With these AI tools getting ever more powerful, everyone has a coworker who's a Steve Wozniak type engineer. You can prompt something to Astra or to Fable and it creates things from scratch. The builder is on call, around the clock, for anyone.
So the difference between two people using the same tool is no longer who can build. It's who can describe what they want. If you can clearly and objectively communicate your vision, your ideas, the things you want built, you get a great product. If you can't, you get a generic one, and you won't know why.
I see this in our own work at VPPA. A lot of what we run for clients sits on top of what we call skills: plain-English instruction files that tell the AI exactly how to do one job. The SEO captions skill we give away is a good example. There's no code in it. It's a few pages of very specific writing about what a good caption is, what to never do, and what the output should look like. The quality of the output tracks the quality of that writing almost one to one. When a skill produces something mediocre, the fix is nearly always a clearer sentence, not a smarter model.
How do you get better at communicating what you want built?
Practice writing the brief before you touch the tool. These are the habits I'd start with:
- Say the outcome first. "A landing page that gets a dentist to book a call" beats "a nice landing page."
- Say who it's for. The same request produces a different (and better) result once the builder knows the audience.
- Say what good looks like. Point at an example. This is where taste comes in: you can only point at good work if you've looked at enough of it to know.
- Say what to never do. Constraints are often clearer than instructions.
- When the result is wrong, say why, specifically. "Make it better" is what a bad manager says to an engineer. It doesn't work on AI either.
Bad: "Write me a caption for this video."
Good: "Write an Instagram caption for this transcript. First line is the question a customer would search, under 125 characters. Second line answers it in one sentence someone could quote. No more than five hashtags, on the last line. Never invent a number that isn't in the transcript."
Same tool, same model. Obviously the second one wins, and nothing about it required engineering.
Who this is for (and who it isn't for)
This is for founders, marketers, creators and operators who were never "technical" and assumed that kept them out of building things. It doesn't anymore. If you can explain what you want to a smart contractor, you can now build a surprising amount yourself.
It isn't an argument that engineers are finished. Knowing how things get built makes your instructions better, and the best results I've seen come from people who have both. It also isn't for anyone hoping the tool will supply the vision. It won't. It rotates shapes.
Common mistakes when working with AI builders
- Prompting before thinking. If you can't write down what you want in a paragraph, the tool can't guess it.
- Blaming the model for a vague brief. Most "AI is mid" complaints are really "my instructions were mid."
- Skipping taste. Communication gets the thing built. Taste tells you whether what came back is any good. You need both.
- Treating it as a software-only idea. Right now this mostly applies to digitally rendered things. I'd bet on it reaching physical things in the near future.
I'm super bullish on this. The better you can communicate what you're trying to build, and the better your taste, the more you'll be able to create great products and solutions and have an impact on the market.
Frequently asked questions
What is a wordcel?
Internet slang for someone with high verbal intelligence, a person who is good with words. It started as an insult. In this post I use it for the articulate person who can express a vision clearly enough for others to build it, a Steve Jobs type.
What is a shape rotator?
Internet slang for someone with strong technical or spatial ability: an engineer, a builder, a person who makes things. Steve Wozniak designing the first Apple computers is the classic example.
Is this just prompt engineering with a different name?
Partly. A prompt is a brief, and a good brief is clear communication. But the skill is wider than prompts. It also covers knowing what you want before you ask, and having the taste to tell a good result from a mediocre one.
Does this only apply to software and digital products?
As of right now it applies mostly to things that can be rendered digitally: code, designs, video, writing. In the near future I expect it to reach all sorts of physical things too.