I’ve been using Claude for about three years. This afternoon, I sat down with a cup of coffee and a head full of ideas. Ninety minutes later, I had a working, interactive 3D instantiation of the Voight-Kampff machine from Blade Runner 2049 live on the internet at voight-kampff.com.
I wrote a PRD in plain English. Claude wrote the code. I drank coffee, gave specific, detailed feedback with common design language, occasionally passed over additional assets, and the machine materialized.
That’s vibe coding. You don’t learn a programming language. You just describe what you see in your head, and the AI translates. But this piece isn’t about the Voight-Kampff machine, or even vibe coding. It’s about what the experience crystallized for me about the future of work—and what every organization should be looking for when hiring humans. (And they should still be hiring humans.)
The Ground Is Already Moving
Earlier this month, Anthropic published “Labor Market Impacts of AI,” measuring not just what AI could theoretically do to jobs, but what it is already doing. Computer programmers sit at 75% task coverage. Customer service reps, data entry clerks, and financial analysts are close behind. No systematic spike in unemployment yet, but hiring of younger workers into exposed roles has quietly begun to slow. The jobs aren’t vanishing. The on-ramp is narrowing.
Meanwhile, StrongDM has published their “Software Factory” framework, and it makes the trajectory unmistakable. Their three-person AI team built a system where humans define intent and then agents take it from there: what the software should do, the scenarios it must handle. The agents generate code, validate it against a “Digital Twin Universe” of cloned third-party services, and iterate until the software converges. No human reviews the code. This is not a thought experiment. It is shipping production security software today.
The implications compound from here. Look at OpenClaw, an open-source personal AI agent framework built by a single developer that rocketed to the top of GitHub in three months with over 260,000 stars. People are using it to deploy swarms of specialized agents that scan industry news, process emails, manage calendars, and execute development tasks in parallel, all running autonomously on their own devices.
In the very near future, it seems clear that instead of browsing and consuming in a typical fashion, a vast segment of consumers and customers will be interacting online via personalized agent swarms like these, each agent operating according to its specialized skillset, acting as an authoritative intermediary to execute any task as requested. Your customer won’t visit your website. Their agent will. Your employee won’t write the code. Their agent swarm will. Even AI search providers like Perplexity will be disintermediated by this shift. When your agent can query, synthesize, and act on information directly, the search layer itself becomes invisible. The entire surface area of digital interaction is being restructured around AI intermediaries, and the humans who direct them will need something that no agent can supply on its own.
1. Creativity: How Well Can New Innovations Be Developed?
To build the Voight-Kampff machine, I didn’t need to know Three.js, WebGL, or JavaScript. Claude is up to speed. What I needed was the idea—the vision of a specific prop rendered in a specific way, with specific behaviors that would make it feel alive. The spark that says “this should exist.” No language model generates that unprompted. StrongDM’s Software Factory still requires humans to define intent. The agents are extraordinary executors, but someone has to dream up what gets executed.
Anthropic’s data shows AI steadily covering the procedural execution layer of knowledge work. What remains uncovered is the ability to imagine something that doesn’t exist yet. Companies that hire for genuine creative vision will be the ones building what the market hasn’t seen, or what niche markets crave.
2. Humanity: How Well Can Innovations Be Communicated?
The hardest part of creating this project wasn’t technical complexity. It was communication: describing what I wanted with enough precision that the AI could build it, while staying flexible enough to adapt when the result wasn’t quite right. That’s not a technical skill. That’s a creative skill. It’s a management skill. And it scales: the same capacity lets you write a compelling project brief, align team work output, or define the scenarios that drive a software factory. In a world of agent swarms, the bottleneck skill is communication; and communication, at its core, is applied empathy. The ability to articulate intent clearly to whoever is receiving it, human or machine.
3. Tenacity: How Hard Does a Person Work at Applying These Qualities?
Ninety minutes sounds fast. But they were ninety minutes of focused, iterative, relentless problem-solving. The first version didn’t look right, so I described what was wrong. The rendering was wonky, the appearance needed refactoring, the position, size, or texturing needed adjustment, so I described the details I wanted. Every cycle was a micro-negotiation between vision and output.
Anthropic’s research captures something important: the gap between AI’s theoretical capability and its actual deployment is enormous. StrongDM’s benchmark tells the same story from the other direction: if you’re not spending $1,000 per engineer per day on tokens, your factory has room to improve. The tools are there. Most people aren’t using them. Creativity without tenacity is a sketchbook that never ships. Communication without tenacity is a perfect brief that never gets built. Tenacity is the multiplier. Tenacity is the ingredient that drives a person to examine a problem from any number of perspectives until it can be solved.
The Takeaway
Stop hiring for tool proficiency—the tools change faster than anyone can track, and the AI is better at them anyway. Start hiring for the person who walks in with an idea that makes you lean forward, who can describe it so precisely you can see it, and who will not stop until it’s real.
This project was constructed with AI tools; the human desire to make something was what made it come to life.
#OmniConsumerProductsCorporation #Qualiaisjob1