In partnership with

This isn't a new story per se. If you've heard the term "vibe coding" in the past two months, you've probably already been exposed to a few interlocking ideas:

1. AI is getting really good

2. It's getting really good at things that are part of our jobs

This has 2 immediate consequences that are part of the current zeitgeist:

1. AI Agent’s increasing capabilities is making people anxious and employers curious

2. It’s ability to take on more work tasks with agency and dependability has unlocked a new kind of work: vibe work.

The first point of AI agent’s getting increasingly capable makes sense to nearly everyone. Even devils' advocates who are paying attention are starting to see the writing form on the walls.

This second point, their ability to do our current jobs is new. This development is special. And probably represents the future of work itself.

Enter Vibe Work 

The core tenent of vibe work is simple: Get good at talking about what you want AI to do. And then let it do it for you.

And I don't mean good. I mean GOOD.

The spirit of vibe work and vibe coding is the next logical chapter in the story of the abstraction of labor.

Knowledge work is no longer a matter of knowing how to manipulate the machinery of ideas; it's about knowing how to effectively communicate with self-manipulating machinery. Having the skills to collaborate and co-create with AI agents to an LLM to get EXACTLY what you're looking for: not just in specificity but wholeness and completeness.

To have a visceral intuition for describing - in terms and references an LLM can work with - what you're looking for.

This new skill is as much project management and creative direction as it is curation. You need to both know how to push a project forward; and how to ask for that thing you're looking for. Human managers will value the skills to communicate in domain specific language that helps shape the intent to your exact specifications. Every profession has its short-hand language to describe complex concepts with concise language. The trick to vibe work is leveraging this short-hand with AI agents.  

Let’s imagine providing art direction notes to an AI agent collaborator: 

Vibe work is as much being a good product manager as it is having a rich vocabulary to talk about what matters to you. It is also relying on increasingly capable AI agents to know the context and short-hand you as the art director using whatever your area of expertise.

And, generally, we call that taste; this is the jumping off point for the new foundation of labor, the taste delta.

What is the Taste Delta?

AI is going to take work.

It's gonna take a lot of work.

So the interesting question is what work - that is to say: valuable task - still remains for the human operator. 

As far as we can tell, the interesting work left to do is all a matter of taste. Narrowly, that the discernment of what is good and what is not good, what is finished and what is not yet quite right, by entirely amorphous, intuitive, and poorly computable criteria is what defines taste. 

The kind of thing that can't really get trapped in training data. 

The kind of thing that can't really get distilled down to a set of describable heuristics we can trap in tokens…

That thing is taste. 

What Is Taste in the Age of AI?

Taste is your human superpower in an AI-driven world. It’s three things:

• Artistic Vocabulary: Mastery of your craft’s language—like "chiaroscuro" in art or "refactoring" in code.

• Domain Literacy: Deep knowledge of your field’s roots and rules, from design history to software patterns.

• Intuitive Judgment: The knack for spotting "good" and knowing when a project’s "done"—like stopping a sketch at its peak.

In vibe work, taste shines through precise, domain-specific language. It’s how you give AI agents clear, token-efficient instructions—just enough words to nail the output.

Taste Jobs Where Humans Lead, AI Follows

Taste defines the future of work across industries. Here’s how humans and AI team up:

• Medicine: A doctor’s empathy guides AI nurses and lab assistants.

• Art & Design: Creative vision shapes AI-generated films, paintings, and layouts.

• Education: A teacher’s coaching pairs with AI tutors for personalized learning.

• Science: Research goals set by scientists steer AI lab techs.

• Finance: Analysts’ strategies come to life via AI bankers and traders.

• Product Management: A PM’s vision drives AI developers and designers.

Humans bring the taste; AI delivers the execution.

The Power of Industry-Specific Language

Every field has its own lingo—jargon that AI agents understand and act on instantly. It’s token-efficient and precise. Examples:

• Design: "Bokeh" tells an AI to blur the background with style.

• Medicine: "Stat" triggers urgent action from an AI nurse.

• Finance: "Short sell" cues an AI trader to bet against a stock.

This succinct language lets veterans direct AI with clarity, turning expertise into results fast. It’s the bridge between human taste and AI capability.

How do we train or learn ‘taste’

• It’s simple we must consume lots of good content that has good taste.

• These deliverables are dependent to profession like we highlighted above. 

Where areas of job growth live

The future of work for humans lies in the taste delta—the gap between what AI excels at and what the market truly desires. We call it the taste delta because it captures the difference between the tasteful output of mechanized systems and the unique flair humans bring. Our thesis? The market will always crave that human touch.

So, what does the taste delta mean for us? Here’s the breakdown:

1. It Shifts by Task

   • For some jobs, like crunching numbers or sorting data, AI delivers results that satisfy. The taste delta here is small—sometimes barely there—leaving room for just a handful of human experts. These are the masters who, after thousands of hours, can spot subtleties AI misses, like a seasoned editor refining a manuscript or a sommelier picking the perfect vintage.

   • In other fields, like crafting a novel or designing a chair, taste is king. The delta yawns wide, and humans shine.

2. It Sparks Creativity

   • Where the taste delta is large, opportunity blooms. In areas like art, music, or product innovation, AI might handle pieces of the process, but it can’t stitch together the full vision. These gaps are where humans step in—co-creating, competing, and infusing their distinct tastes into work that resonates.

   • Think of a filmmaker tweaking a scene or a chef balancing flavors—tasks where the human edge turns good into extraordinary.

The taste delta isn’t just a niche—it’s the frontier where human ingenuity thrives. As AI handles the predictable, our ability to surprise, delight, and innovate will keep us in demand. The market’s appetite for taste ensures it.

What synthetic labor is missing

Today, AI agents know these taste derived creative tasks as the negative space. Things that don’t tokenize well. All the labor that ai agents cannot do, but the market demands. 

Human agency will flock to the tasks that the market demands, but AI agents cannot do.

We have taken our first step into defining the taste delta, by highlighting what is missing. The next step is to define what that negative space needs to be filled with to meet the market needs.

1. Today’s AI labor - AI agents are the worst they will ever be today and they’re getting better rapidly. There are big gaps in its capabilities and really thoughtful people have empirically tested these gaps, see: Navigating the Jagged Technological Frontier.

2. AI agent labor - With synthetic labor we can create near limitless explorations and research topics. The labor is consistent, but has consistent skills gaps it is simply unable to fill.

3. Filling the taste gaps - These gaps are areas of creativity, taste and decisions of the definition of done for a task.

4. The hybrid economy -  These tasks are executive functions to direct and drive labor to successful completion. 

The birth of this new human-ai hybrid economy will be messy, but has the potential to unlock new opportunities. It is our job as innovators and creatives to begin finding and defining these new roles and opportunities for human ingenuity.

Mapping the taste delta in your profession

It starts with understanding where your taste—your ability to vibe code, art direct, and say "that’s the one"—sets you apart. Here’s a framework to assess your taste gap and build a career that’s not just future-proof but future-defining.

1. Map Your Taste Territory

• What’s your edge? Pinpoint the skills in your toolkit that AI can’t touch. Maybe it’s your knack for spotting when a UI feels "off," your intuition for pacing a user flow, or your ability to dream up a feature that no dataset could predict. These are your taste-driven superpowers.

• Speak the language: Your fluency in domain-specific shorthand—like "glassmorphism" or "bokeh" in design, or "async await" in coding with a creative twist—lets you steer AI agents with precision. Taste isn’t just feeling; it’s articulating what you feel in terms an AI can quickly implement.

• Define "done": AI can iterate forever, but you know when a project sings—when the landing page hero shot lands just right or the code hits that sweet spot of elegance and utility. That’s taste in action. How will you find the places in your work process that allow you to uniquely position yourself to critique, refine and complete work?  

Action: Jot down three things you do in your work that rely on your gut, your eye, or your ear—stuff AI can’t fake. That’s your taste territory.

2. Spot the Taste Gaps in Your Craft

• Where does AI stumble? Look at your field and ask: where does AI nail the grunt work but miss the soul? In design, it might crank out layouts but flub the emotional arc of the experience. In development, it might write functional code but lack the flair that makes an app addictive.

• What does the market hunger for? Clients and users don’t just want functional—they want delightful, surprising, human. Think about what keeps your work in demand: the clever UX tweak, the bold aesthetic choice, the feature that sparks joy. That’s the taste gap AI agents can’t fill.

• Size the delta: Some tasks (like debugging boilerplate code) have a tiny taste delta—AI’s got it covered. Others (like crafting a brand’s digital voice) have a gaping delta where your taste reigns supreme.

Action: Pick two tasks you do regularly where AI can assist but falls short of perfection. These are your taste gaps—own them.

3. Build Your Taste-Driven Career

• Chase the high-taste roles: Focus on gigs where taste is king—think product vision, creative direction, or experience design. Let AI handle the pixels and the loops while you set the North Star. Picture yourself as the product manager vibing with an AI agent dev team, or the art director guiding an AI designer to brilliance.

• Team up with AI: Don’t fight the machines—vibe with them. Use AI agents to mock up ten hero shots or spin up a prototype, then swoop in with your taste to pick the winner and polish it to perfection.

• Keep sharpening your taste: Consume the good stuff—great designs, killer apps, boundary-pushing code. Taste grows by osmosis, and the more you soak in, the wider your delta gets.

Action: Choose one way to offload a low-taste task to AI this week (e.g., generating initial wireframes). Then, set a goal to level up one high-taste skill in the next three months. Ask experts about the most impactful books and sources they consume. Dig deep into the history and the culture and the latest trends of your industry, so you can seamlessly use the short-hand language to vibe work with your AI agent team. For UX design use the latest apps, decompose the best work on Mobbin and read foundational work like Alan Cooper’s About Face. 

The market’s waiting—and it’s got an appetite for what only you can bring.

About My Guest Co-author This Week

Start learning AI in 2025

Everyone talks about AI, but no one has the time to learn it. So, we found the easiest way to learn AI in as little time as possible: The Rundown AI.

It's a free AI newsletter that keeps you up-to-date on the latest AI news, and teaches you how to apply it in just 5 minutes a day.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses – tailored to your needs.

Sign up to start learning.

🧲 AI Agent Magnet

Anthropic just released Fleur their open-source app-store for MCP, Model Control Protocol, based apps that allow agents to use tools. The App Store allows users to easily add MCP tool apps. The Anthropic team had noticed there is friction to using MCP as it requires coding experience.

Inception Labs just created, Mercury, the fist of its kinds diffusion LLM to dramatically accelerate text generation. Instead of 75+ iterations per output with existing LLMs, it delivers refined answers in around 14 iterations. The flagship diffusion model achieves over 1000 tokens/second on standard Nvidia H100 GPUs. No custom hardware required—this performance leap is accessible immediately.

A new startup serving as a tool to design apps from text prompt to interactive prototype. The tool is useful in quickly outlining ideas for rapid prototyping and testing. I’m interested to see this and other AI agent design tools evolve. This early app needs guiding in future iterations it will prompt the user with questions to build a scope and then create the prototype. I share this link to try it out and keep an eye on how AI agent designers evolve.

💬 Suggestion box

A newsletter exploring the principles that will help us design AI agent experiences and startups that amplify human creativity.

Subscribe to join a community of designers and developers shaping purposeful AI agents.

Referral Rewards

Until next time, keep innovating and stay curious!

Jason