AI is the Engine that Can Build Anything. Design Is Steering Wheel. The Direction Turns Output Into Impact.

These days, everyone's so excited to show me the website or app they just made, and I genuinely love that so many more people can build now. That part is beautiful. But I also feel the rising tide of noise and the lack of mindfulness underneath it all.

As AI takes the building, a designer’s job becomes making the experience deeply human, and guiding everyone else to build it right. Less pushing pixels. More creating, pushing for, and holding the soul of the experience while the team moves fast around it.

Building products is a great responsibility. What we ship changes how people feel, think, and act. It touches their security, their identity, their productivity. After nearly a decade building products used by hundreds of millions of people at Microsoft, and founding the work that extended inclusive design into mental health and cognition, I've seen the unintentional harm that comes from careless, fast product making. A default that quietly leaves someone out. Notifications meant to help that only add stress. Settings that drive people crazy. A flow that assumes everyone thinks exactly like the person who built it. Anything can be created now. Only the products with real design and real systems beneath them will thrive.

So what is good design? Most people picture fonts, colors, the look of a product. That matters, but it's the most visible piece of a much bigger puzzle. When I told people I worked on Outlook, they'd ask, "So do you pick the colors on the calendar?" and I'd laugh. Design isn't making things pretty, and I love blowing people's minds with what it actually is. It’s closing the distance between what a product is, what a person needs, and how they move through it. That has never mattered more than right now, because when anyone can build almost anything, the only question left is whether we're building the right thing.

I chose game design over graphic design because I was obsessed with human behavior and how technology helps people level up. To feel confident, capable, and empowered beyond their existing limitations. That’s still my definition of good design. It points a business and its people in the same direction, so they win together.

What people count as winning has changed. Research has made it unmistakable. Productivity no longer means maximizing efficiency. Look up productivity in the App Store. It’s apps that help with focus, routine, and calm. People don’t want to do a million things. They want to feel like they’re winning in an economy built to steal their attention. The feeling of success hits harder than the sum of it.

Think about the best product you’ve ever used. It feels simple. Natural. You use it without thinking, because it knows you and celebrates you. You love the soul behind it, because it helps you become a better version of yourself. What made that happen? Humans who know humans. The magicians behind the interface.

AI is incredible at executing, generating, adding. It cannot decide what should matter, what deserves to be celebrated, what the soul of a product should be. That is the foundation. Without it, you’re only driving faster in the wrong direction, burning fuel, lengthening the road home. With it, creation unlocks. Ferrari mode. AI is the engine that can build anything. Design is the steering wheel.

Building with AI is like working with any great partner. You get great together by being honest about what each of you does best. A collaborator, not a replacement. Before I hand something off, I ask one question. Does this feel like math? Like execution? If it does, it’s AI’s. Then I fill in the gaps with the human parts, the taste, the judgment, the care. As AI gets stronger at implementation, your leverage moves upstream. Your judgment belongs before, during, and after the work, iterating until it feels right. AI carries you far towards a screen, but a screen is not an experience, and I watch people confuse the two every day. Once you’re live, trust takes a second to lose and forever to earn back. Flip it around. Make someone feel understood in the first moment, help them reach a new level they couldn’t touch alone, and that trust and product love compounds fast.

Helping teams find and love their direction

While I was recently in Japan, I learned a word that became my foundational design principle. Kokoro: the alignment of heart, mind, and spirit.

That's what I try to help every project reach. The heart is what they're trying to do and why it deserves to exist. The mind is the strategy and system behind how it works. The spirit is how their users feel it in practice. Once that alignment clicks, everything accelerates in the right direction, and only then do we build the arc and the systems AI can genuinely execute against. Here's a small preview of how I get a team there. It shifts based on the goal, the deliverable, and the context.

Define, write the hero's journey. First we ground ourselves in what we actually believe as a company, and get honest about whether what we ship reflects it. We align on the gaps, outcomes, constraints, research, and what "wow, this is it" means for both users and the business. Then, before any screens, we write it like a movie script, whether it's a whole product or a single feature. The story is simple: users are struggling, they start using it, and their lives get meaningfully better. By the final scene everyone wins, including the back of house that runs the business, and everyone building it knows exactly where they fit. Heads up, the user and the paying customer are often different people, so map everyone involved and serve both. This gives the team one shared direction before we waste credits and energy building the wrong thing.

Align, create and evangelize a practical vision. Now the story shifts from movie into a game. What does each person need to do, how do they level up to that winning feeling, and how can the experience carry them straight through their most tedious, time-consuming pain points? How can we create rainbow roads for users to zoom through their existing flows? Recognizing the little wins along the way is the gas pedal on flow. It creates feelings of capability and success, and bonds people to the product, because everyone loves to feel progress and be celebrated. We meet users where they are and let the experience adapt to their definition of success, context, and urgency. This is where we get tactical about the phases, multimodal inputs, mindsets, and actions of everyone involved. What comes out the other side is a vision aligned with the core principles that the team, the investors, and the customers can't wait to build, from the MVP to phase ∞.

Execute, full throttle, but on rails. This is the payoff of all that upstream work. You finally know exactly what you’re building, why, who needs it, and when, so the building goes fast. The speed is only safe because the vision already laid the track. Execution runs full throttle, but inside systems and guardrails the vision drove. This is where agentic tools shine. A clean component library is table stakes now. The harder part is the web of rules stacked on top of the system, rules we’re all still learning to write. What matters most? What can move? What must never move? What should the thing feel like in different moments, how should it bend per person, what happens when it isn’t sure? A clear story and a defined system are the whole difference between AI amplifying your direction and confidently wandering away from it providing clarity before and quality assurance after. For example, I love that we’re able to systemically say “include usability and accessibility best practices” and it’s auto applied to the team that’s building. That’s what keeps the throttle safe and in line.

Upward spiral, experiment, learn, rise. I call it a spiral because every experiment should be generative, each turn should lift the product higher. Recognize what a user does, reflect it back, reward their progress, personalize, deepen trust, and repeat. TThe magic isn’t in adding more. It’s in listening. You will never help complexity reach simplicity by adding to it. Teams get obsessed with shipping features when the real work is quiet. Reading the data, hearing what people are actually telling you, moving on what matters. Let’s also please stop measuring success by engagement alone, if someone hits their goal in less time that's a win, not a reason to keep them hooked. Numbers only tell half the story, so pair them with how people actually feel. The business outcome and the human experience will reinforce each other and grow together.

How I partner with AI

I'm surfing the waves of change with everyone else, so I'm always refining this, but a few things are really working for me right now.

Push your input as far as you can, then hand off. Words aren’t the only way to prompt. Working straight in Figma, or pointing at the exact part of an experience I want to nail, carries me further and faster than describing it in text, and it keeps me in flow. When I want something to feel original, or perfect, I take it as far as I can by hand, then give AI a draft to clean up. With the Figma MCP I move back and forth. Draft, react, refine, repeat. AI is unpredictable by nature, so always make sure to have the final approval.

Apply and iterate the system, don't start from scratch. Don’t start from scratch. When a design system is built for AI, with clean components, clear rules, real context, I’ve watched people who’ve never called themselves designers make useful, on brand work. A strong designer makes the whole team faster and cleaner even when she’s not in the room. I loved watching a recent client light up at what he and his team could spin up once he connected his system to AI and now watching him run. Strong systems already exist. Adopt one that works well, then make it uniquely yours.

Let AI be specialized direct reports, not the director. AI has become a great critique partner and teacher, especially on what I don't know. I use it for tedious work, consistency checks, exploring approaches, and figuring out which model or workflow fits a problem. I build custom GPTs filled with knowledge, and I love using Claude's plan mode to review the approach before letting it touch the work. Nothing it hands me is the final answer, it's the start of a conversation.

Synthesize, structure, systematize. I love yapping into WisprFlow, spilling every half formed thought, and letting AI find the shape hiding inside it. It turns thinking out loud into a system, and from there into clear models for architecture and tradeoffs, and agents with real rules and guardrails to run.

Staying uniquely human

Left unguided, AI pulls toward the probable, the familiar, the middle of everything it's ever seen. That middle is the enemy. The easier it becomes to make something, the more sameness floods the world, and the more a real point of view is worth and will drive business outcomes. Average is exactly what the machine will hand you the moment you don’t push for more and build with true care and intention. Personalization and product truth are your only real differentiation. They’re what make people feel seen, stay, and fall in love.

Look at Spotify. One platform that somehow feels handmade for every single person inside it, all in service of one love, finding and falling for music. The recommendations, the layout, the Wrapped ritual every December, the whole system flexes until it feels like the most me. That is recognition and delight at scale.

So before you push a single pixel, connect to what you’re building in the ways only you can, or hire designers like us to help you do so. Get clear on how it lifts your whole spectrum of people while moving the business forward. Find that alignment, lay that foundation, then let AI accelerate it. In a world that will become full of noise and slop, the deeply human thing is the one people will relate to and actually fall in love with.

‍

Ali Benter, a former Microsoft design leader across Copilot, Outlook, and Inclusive Design for Mental Health & Cognition, now helping companies accelerate through independent advisory work and Flexxi Studio, an embedded partner for AI, product, brand, and web.