By ยท

The Human Skills That Will Still Matter When AI Can Do Almost Everything

Every conversation about AI and work eventually lands on the same anxious question: what happens to us? I get it. The pace of change is real, and the list of workflows being touched by AI right now is not short. Copywriting, prototyping, programming, research, planning, image generation, video, sound. At varying levels of maturity, yes, but all of them in motion. If you work in tech or digital media, you have already felt this.

But I want to offer a different frame before we spiral into the usual dread. The more interesting question is not what AI is replacing. It is what it is revealing. Because as AI starts absorbing the how of work, the things left standing are the parts that were always hardest to systematize. The parts that are distinctly, stubbornly human.

Here is what I actually think is happening, and what it means for anyone building a career or a product right now.

AI Is Giving Everyone a Junior Collaborator

The most honest way I can describe what is happening across tech product companies is this: AI is functioning like a capable, always-available junior person sitting next to every individual contributor. A PM can now draft briefs, run competitive research, and generate early concepts faster than ever. A designer can move from vague idea to working prototype in a fraction of the time. A developer can get unstuck on a problem without waiting for a code review.

That is not a small thing. The productivity unlock is real, and it is not evenly distributed yet, which means the people who learn to work with these tools effectively are already operating at a different level than those who have not.

But here is the part that gets overlooked in the hype: having a capable junior collaborator does not eliminate the need for a senior judgment call. It actually makes that judgment more important. When output gets faster and cheaper, the quality of direction becomes the real differentiator.

Intent Is the New Core Skill

If I had to name the single most critical human skill in an AI-augmented workflow, it would be intent. Specifically, the ability to decide what you want and why you want it, before you ever touch a tool.

Prompts are the primary interface we have with most AI models and agents today. And the quality of a prompt is almost entirely a function of how clearly you can articulate your intent. You cannot delegate your way to a good outcome if you do not know what a good outcome looks like. The AI will fill in the gaps, and it will fill them with averages, with the most statistically common version of what you asked for.

Intent Is the New Core Skill If I had to name the single most critical human skill in an AI-augmented workflow, it would be intent. Specifically, the ability to decide what you want and why you want it, before you ever touch a tool. Prompts are the primary interface we have with most AI models and agents today. And the quality of a prompt is almost entirely a function of how clearly you can articulate your intent. You cannot delegate your way to a good outcome if you do not know what a good outcome looks like. The AI will fill in the gaps, and it will fill them with averages, with the most statistically common version of what you asked for. This is why I think decision-making clarity is becoming a premium skill. Not just for product managers or founders, but for anyone operating with AI tools. The how is increasingly handled. The what and the why are yours to own.

This is why I think decision-making clarity is becoming a premium skill. Not just for product managers or founders, but for anyone operating with AI tools. The how is increasingly handled. The what and the why are yours to own.

Orchestration: Managing More Than One Thing at Once

The next layer is orchestration, and I think this one is underrated in most conversations about AI and work.

What I mean by orchestration is the ability to work with multiple AI models or agents, sometimes sequentially, sometimes in parallel, and get them to produce something coherent together. Think about a workflow like research, concepting, and design. You could pass outputs from one stage to the next across different tools, or run certain tasks simultaneously and then synthesize them. The person who can set that up and direct it is operating more like a conductor than a contributor.

This does not require deep technical knowledge. It requires systems thinking, some comfort with ambiguity, and a clear sense of what you are trying to build. Those are judgment skills, not purely technical ones. And as AI agents become more capable, the ability to orchestrate them well is going to separate people who get leverage from people who just get busy.

Craft and Taste Are Not Going Anywhere

This is where I want to push back on a narrative that has been floating around, which is that AI is developing taste. I have used enough tools at this point to say: not really.

What AI has is an enormous library of design systems, visual references, published frameworks, and documented conventions. When you ask it to make something look good, it will pull from that library and produce something that looks like a competent average of the things it was trained on. It defers to the familiar. It gravitates toward what has already been validated.

Craft and Taste Are Not Going Anywhere This is where I want to push back on a narrative that has been floating around, which is that AI is developing taste. I have used enough tools at this point to say: not really. What AI has is an enormous library of design systems, visual references, published frameworks, and documented conventions. When you ask it to make something look good, it will pull from that library and produce something that looks like a competent average of the things it was trained on. It defers to the familiar. It gravitates toward what has already been validated. That is useful. It is genuinely useful. But it is not taste. Taste involves knowing when to break the pattern. It involves having a point of view that is not reducible to what already exists. It involves the kind of creative instinct that comes from years of looking at work, forming opinions, and making things that did not have to exist. Craft and taste are human. They are also becoming rarer in a world where it is very easy to produce something that looks finished. The ability to tell the difference between a thing that is technically correct and a thing that is genuinely good is going to matter more, not less, as AI makes technical correctness cheap.

That is useful. It is genuinely useful. But it is not taste. Taste involves knowing when to break the pattern. It involves having a point of view that is not reducible to what already exists. It involves the kind of creative instinct that comes from years of looking at work, forming opinions, and making things that did not have to exist.

Craft and taste are human. They are also becoming rarer in a world where it is very easy to produce something that looks finished. The ability to tell the difference between a thing that is technically correct and a thing that is genuinely good is going to matter more, not less, as AI makes technical correctness cheap.

What This Means If You Are Building a Career Right Now

I was recently at a panel on AI, job search, and digital media futures, and a lot of what I kept coming back to was the portfolio question. If you are a designer or a developer entering the market right now, your portfolio is not optional. It is your proof of intent, craft, and judgment all at once.

My advice: do not try to show everything. Pick your best one or two pieces and go deep. Walk through the process. Show the problem statement, the thinking, the decisions you made and why. Less is more. One well-documented project beats five shallow ones every time.

Beyond the portfolio, networking still matters, and I do not mean that in the vague, go-to-a-conference sense. I mean hackathons, meetups, finding the places where the people you want to work with actually spend time. Relationships built in those rooms tend to be more durable than ones that start with a cold application.

What This Means If You Are Building a Career Right Now I was recently at a panel on AI, job search, and digital media futures, and a lot of what I kept coming back to was the portfolio question. If you are a designer or a developer entering the market right now, your portfolio is not optional. It is your proof of intent, craft, and judgment all at once. My advice: do not try to show everything. Pick your best one or two pieces and go deep. Walk through the process. Show the problem statement, the thinking, the decisions you made and why. Less is more. One well-documented project beats five shallow ones every time. Beyond the portfolio, networking still matters, and I do not mean that in the vague, go-to-a-conference sense. I mean hackathons, meetups, finding the places where the people you want to work with actually spend time. Relationships built in those rooms tend to be more durable than ones that start with a cold application. On the role evolution side, a couple of things are worth paying attention to. There is a new category of contributor emerging that some teams are calling forward-deployed designers and engineers. The idea is that the individual contributor goes directly to the client, does the discovery, comes back with concepts, and iterates without needing a product manager in the middle to translate. This is still early, and it is mostly showing up in smaller teams and early-stage startups. But it points to something real: client-facing communication is becoming a core skill, not just a nice-to-have. Product managers, meanwhile, are shifting toward something closer to product marketing. The ability to document what a feature does, create a walkthrough, and find ways to communicate its value clearly is becoming part of the role in ways it was not a few years ago. If you are thinking about career progression in product, start building that muscle now.

On the role evolution side, a couple of things are worth paying attention to. There is a new category of contributor emerging that some teams are calling forward-deployed designers and engineers. The idea is that the individual contributor goes directly to the client, does the discovery, comes back with concepts, and iterates without needing a product manager in the middle to translate. This is still early, and it is mostly showing up in smaller teams and early-stage startups. But it points to something real: client-facing communication is becoming a core skill, not just a nice-to-have.

Product managers, meanwhile, are shifting toward something closer to product marketing. The ability to document what a feature does, create a walkthrough, and find ways to communicate its value clearly is becoming part of the role in ways it was not a few years ago. If you are thinking about career progression in product, start building that muscle now.

The Longer View

I want to be honest about the longer arc here. The roles we are talking about today may look quite different in ten years. AI is currently helping with chunks of workflow. Eventually, in specific verticals and with enough training data, it may get good enough to handle entire roles end to end. That is not alarmist. It is the natural pattern of how innovation displaces and creates work, a cycle that has played out across every major technological shift.

But two to five years out, which is the timeline that actually matters for most of the decisions people are making right now, the picture is different. The skills I keep coming back to are the same ones I already mentioned: clear intent, orchestration ability, and genuine craft and taste. Plus one more that I would add for anyone thinking about long-term positioning: visibility.

The Longer View I want to be honest about the longer arc here. The roles we are talking about today may look quite different in ten years. AI is currently helping with chunks of workflow. Eventually, in specific verticals and with enough training data, it may get good enough to handle entire roles end to end. That is not alarmist. It is the natural pattern of how innovation displaces and creates work, a cycle that has played out across every major technological shift. But two to five years out, which is the timeline that actually matters for most of the decisions people are making right now, the picture is different. The skills I keep coming back to are the same ones I already mentioned: clear intent, orchestration ability, and genuine craft and taste. Plus one more that I would add for anyone thinking about long-term positioning: visibility. Start sharing your thinking. Not for immediate engagement, not to go viral, but to build a trail of ideas that represents who you are and how you think. Agents will increasingly be doing background research on people before anyone in a hiring or investing conversation even reaches out. The people with a longer, richer knowledge trail are going to carry more weight when that moment arrives. The tools are changing fast. The judgment, the taste, and the clarity of intent behind how you use them are still yours to develop. That is worth taking seriously.

Start sharing your thinking. Not for immediate engagement, not to go viral, but to build a trail of ideas that represents who you are and how you think. Agents will increasingly be doing background research on people before anyone in a hiring or investing conversation even reaches out. The people with a longer, richer knowledge trail are going to carry more weight when that moment arrives.

The tools are changing fast. The judgment, the taste, and the clarity of intent behind how you use them are still yours to develop. That is worth taking seriously.