The world of software design is changing fast ​

For a long time, building software products followed a familiar path. A team would do research, sketch out rough ideas, and then turn those ideas into polished designs inside Figma. That tool became the center of everything. It was the "source of truth," and every decision flowed through it for our agency Border UX.

We don't see it that way anymore.

We believe software design and development is becoming automated. Over the next few years, more and more of the work that used to take large teams of people will be done by AI and other smart tools. That doesn't scare us. In fact, it’s driving us toward a new version of our company. Instead of fighting that change, we decided to shape it.

We flipped the old model. Figma is no longer the center of our universe. It's just one place where the finished work shows up, the same way a web browser shows a finished website. The real work now happens somewhere new.

Why raw AI isn't enough ​

You can hand a problem to a tool like Claude Design, ChatGPT, or Gemini today and watch it build a screen in minutes. That feels like magic. But there's a catch.

AI has learned from millions of examples across the internet. Some of those examples are good. Many are not. Because it has seen everything, it thinks anything is allowed. There's no single "right way" baked in. It tends to do one of two things. It either creates strange, messy designs that don't make sense, or it plays it so safe that everything looks the same and boring.

The bigger problem is this: AI doesn't know when to stop. It tries to give you everything at once because it's hoping that will make you happy. But that's not what makes a product good. A good product is one that is finished, correct (based on the information at hand), and easy for a real person to use.

That's the gap we fill.

We give AI good taste and firm rules ​

We've spent decades designing and building software. Over those years we learned which patterns work, which ones fail, and why. We learned the difference between a screen that looks fine and a screen that people can actually use. We know how to use human psychology and patterns to make something easy to use.

Now we take all of that hard-won knowledge and pour it into our tools. We don't just write better instructions for the AI. We build real guardrails. We give it a toolbox and tell it that it can't go outside that toolbox. That way, we don't have to cross our fingers and hope the AI gets it right. We make sure it does.

Think of it as layers of control. There are colors, styles, design tokens. There are reusable building blocks, like buttons, menus, and panels. And on top of those, there are rules about how and when to use each piece. Each layer adds a little more guidance. Together, they push the AI to build the way an experienced human would build.

This is what some people call "taste." Every product has a style and a feel. Once you give the AI that taste and those limits, it can move fast and still produce something that feels real and professional. That's the difference between our approach and simply opening a chatbot and hoping for the best.

Built for the products people forget about ​

We focus on enterprise software. That means the tools companies use inside their own walls, often by their own employees.

These products matter, but they're often a mess. Many were thrown together quickly by a few engineers with no designer involved. They're packed with data and hard to use. The good news is that fixing them usually doesn't require wild creativity. It just requires following solid, proven practices. When you do that, the experience gets dramatically better.

Our goal is to make those best practices the default, so good design happens almost automatically.

Your work stays yours ​

We open-source our tools and let companies run them on their own systems. There's a reason for that.

When a company does research or builds a product, it's handling sensitive information. Sending all of that to an outside SaaS product or service is a real risk. We don't want to hold anyone's private data, and we don't think anyone should have to hand it over just to use good tools. We built our approach to avoid that problem completely. You're never locked into a subscription, and your secrets stay your own.

A new role for people ​

As tools get smarter, the role of people changes too.

In the past, you needed a designer to push pixels and an engineer to write code before anyone could see an idea come to life. Soon, the person who actually understands the problem can talk directly to the technology and watch a real product appear. They won't need to know design tools or coding languages. They'll just need to know what they're trying to solve.

The catch is that, on its own, that fast result is often low quality. Our tools take care of the parts stakeholders can't see, like usability and consistency. The finished work is something they can trust.

What we believe makes design good ​

A few simple ideas guide everything we build.

We believe in understanding how people actually work, step by step, before deciding what to build. We believe that people live in a physical world, so digital things should behave in ways our brains already understand. And we believe in simplicity, which means knowing what to leave out. Editing is the hardest part, and it's the part AI struggles with most.

In the end, quality comes from one thing: knowing the goal, then reaching it in the best way you can with what you know at the time.

Where we're headed ​

We're not hiding from AI. We're handing it everything we've learned so it can work better and faster for the people who use it.

Our job is no longer to do all the hands-on labor. Our job is to bottle up our knowledge into tools and systems, keep making those tools smarter, and guide the whole process. If we can shorten the distance between "here's my problem" and "here's a high-quality product," then we've done what we set out to do.

That's the future we're building.