Designing with AI: What I Am Actually Trying and What I Am Learning

I am not waiting for permission to understand AI.

I did not want to be the designer who had opinions about AI without ever actually using it. So I started building. This is what I am learning across my design practice, inside product work and in the personal projects I run on my own time with no stakeholder in sight.

My honest take: AI has not made design easier. It has raised the bar for what good design thinking actually looks like. The skill that matters most right now is not knowing the tools. It is knowing how to think through them.

The designer as integration layer

The conversation about AI in design tends to default to replacement anxiety. I do not think that is the interesting question. The interesting question is what it means to design when the tools themselves are becoming intelligent.

My answer is that the designer's role is evolving into something more like an integration layer. Someone who understands enough about user behavior, enough about how AI systems work, and enough about what code can actually do to connect systems that do not yet know how to talk to each other.

That is a different job than the one most of us were hired into. It requires fluency that is not just visual or strategic. It requires genuine curiosity about the underlying systems, a willingness to experiment, and comfort with not having a clean answer every time I sit down to work. The designers who will thrive are not the ones who found the perfect stack. They are the ones who stay in motion as things shift and keep building regardless. I am trying to be one of those designers.

Being an AI-first designer means starting earlier

I do not wait for a research readout or a full brief to start forming a point of view. I use AI to synthesize early signals, surface patterns across user insights, and pressure-test flow logic before a single frame is open. By the time I am in Figma I already have a hypothesis worth designing toward.

What that changes most is the quality of my discernment. When I can generate multiple directions quickly, my energy goes into deciding which one is actually worth pursuing rather than scrambling to produce options. That judgment, knowing what good looks like and why, is where the real value of a senior designer lives right now.

The making is faster too. I prototype concepts early and test them before the conversation moves into execution. And I stay genuinely curious about what is possible at the edges, playing with interaction patterns and moments of delight in ways that feed back into better design decisions upstream.

Prompting is a design presentation skill

How I prompt an AI system for a prototype under strict context is not that different from how I write a product brief or present design work. Both require clarity of intent, to anticipate edge cases, to explain the product to someone who has never interacted with it before. It requires me to know when the output is good enough and when I need to go again.

Prompting is not a trick. It is a communication skill, and it is the same one that makes me a better designer: articulating a problem precisely, providing the right context, and knowing when to push back on what I get back.

We are moving from AI as a tool I query to AI as a system with real skills it can execute within a workflow. Knowing how to direct that, scope it, and verify the output is going to be one of the defining competencies for designers over the next few years. It maps directly onto what good product thinking has always been: knowing what problem I am actually trying to solve, building something that addresses it, and being honest about what is working and what is not.

Figma as a thinking tool, not just a design tool

When I feed a complex problem into Figma Make and watch how it interprets the visual logic, the output is usually off unless I prompt strategically. But the way it fails tells me something. It forces me to articulate the problem more precisely, and that act of precision is where the real design thinking happens. When the system is not producing the right output, the question I ask myself is what context is missing. That gap is almost always where the sharper design decision lives.

The most valuable thing AI does in my design process is compress the time between a fuzzy problem and a sharp hypothesis. The judgment about which hypothesis is worth pursuing stays with me.

Building to think more clearly

Some of the most important learning I am doing is not happening inside a product team. It is happening in the hours I carve out to build things for myself, personal projects with no brief, no stakeholder, and no approval process. That freedom is where I am developing the most transferable fluency.

My personal stack is Claude Code with the Figma MCP, Lovable, Replit, and Vercel. Lovable gets something running fast so I can test a product idea at the pace I think. Replit is where I prototype and experiment without friction. Once a concept holds up I rebuild it in Claude Code for more precise control over the translation between design intent and actual implementation, and I deploy with Vercel. Shipping something, even a small personal project, changes how you think about design in ways that staying inside a design tool simply cannot.

I have also been exploring Antigravity, which has pushed how I think about interaction and motion in ways I am still processing. It is the kind of tool that does not fit neatly into a workflow yet but is worth staying close to.

When I actually build things, even imperfect ones, I develop a feel for feasibility that wireframes alone cannot give me. I ask different questions in design reviews. I understand where the complexity actually lives. Working daily alongside engineers at Volvo, an engineering-first company, has deepened that further. It is giving me a real appreciation for what it means to push code, navigate security considerations, and think about scalability as a design constraint worth understanding from the start.

What I am still figuring out

The tools I use today may not be the ones that matter in six months. What I am more confident about is the posture. Stay curious. Build things. Do not wait for permission to understand the systems shaping your industry.

That is the practice. The tools are just the current expression of it.