I should probably start with a confession. The first version of Kaer was terrible. Genuinely, embarrassingly bad. It generated websites that looked like they'd been designed by someone who had never actually seen a website before. Layouts that made no sense. Colour combinations that could only be described as aggressive. Navigation menus that led nowhere.
I remember showing an early prototype to a friend, someone whose opinion I really valued. He looked at the output for a long moment, then looked back at me with this expression that was trying so hard to be supportive. "It's... interesting," he said. "Is it supposed to look like that?"
It wasn't. But that conversation, as deflating as it was in the moment, crystallised something important for me. I wasn't just trying to build another AI tool. I was trying to solve a problem that had bothered me for years.
The Frustration That Started Everything
Here's the thing about building websites. It's simultaneously too easy and too hard. The technical barriers have never been lower. You can spin up a site in minutes using any number of platforms and templates. But creating something that actually feels right? Something that captures what you're trying to say, that resonates with the people you're trying to reach? That still requires either genuine skill or significant money.
I've watched people struggle with this disconnect countless times. Friends starting small businesses who needed a web presence but couldn't afford a professional designer. Creatives with brilliant ideas trapped behind generic templates that stripped away everything that made their work distinctive. People spending weeks wrestling with website builders, trying to force their vision through interfaces that were designed for someone else's vision entirely.
And I kept thinking: this doesn't have to be so hard. The gap between "I know exactly what I want" and "I have exactly what I want" shouldn't require technical expertise or design training to cross. What if there was something in the middle? Something that could understand intent and translate it directly into reality?
The Leap of Faith
When large language models started demonstrating genuine capabilities in creative tasks, I saw an opportunity. These systems could write coherent text, understand nuance and context, even engage with aesthetic concepts in surprisingly sophisticated ways. The question that consumed me was whether they could be guided to not just generate content, but to make the kinds of holistic decisions that good design requires.
I won't pretend I knew what I was doing when I started. My background was more technical than creative, and the intersection of AI and design was uncharted territory. I was learning on the fly, making mistakes constantly, hitting walls that seemed insurmountable until suddenly they weren't.
The breakthrough — if you can call it that — came when I stopped thinking about the problem as "get an AI to write website code" and started thinking about it as "teach an AI to think like a designer." Those are fundamentally different challenges. The first is a technical problem. The second is almost philosophical. What does it mean to make good design decisions? How do you balance aesthetics with function? When do you follow conventions and when do you break them?
I spent weeks just studying how designers work. Not the tools they use, but the thinking processes behind their choices. The way they iterate, refine, step back and reassess. The questions they ask themselves. The intuitions they develop over years of practice. I wanted to distill some of that into something an AI could learn from.
What Kaer Actually Does
At its core, Kaer is an attempt to build an AI that can be a genuine creative partner in web creation. You describe what you want — in plain language, as vaguely or specifically as you like — and Kaer works to understand not just the words you're saying, but what you're actually trying to achieve.
This goes beyond simple prompt-to-output generation. Kaer asks clarifying questions when your intent is ambiguous. It makes suggestions you might not have considered. It explains its reasoning so you can guide it better. And crucially, it iterates. The first version is never the final version. You respond, it adapts, and gradually you converge on something that actually represents your vision.
The technical architecture behind this is something I'm genuinely proud of, though I'll spare you the details here. What matters more is the philosophy: treat the AI not as a tool that executes commands, but as a collaborator that brings its own capabilities to the partnership. You provide the vision and the judgment. It provides the execution and the exploration of possibilities you might never have imagined.
The Bigger Picture
Web creation is just where I'm starting. When I think about where agentic AI is heading, I see a world where the barriers between imagination and creation collapse entirely. Not just for websites, but for software, for physical products, for experiences of all kinds.
Right now, there's an enormous amount of human potential locked away behind technical barriers. People with brilliant ideas who can't code. Visionaries who can't design. Entrepreneurs who can't afford to hire the specialists they need. Every one of those barriers is an opportunity for AI to democratise capability.
Kaer is my contribution to that larger project. It's small, focused, imperfect. But it's real. It works. And every day I'm making it better, learning more about what's possible, pushing into territory that didn't exist a few years ago.
Why I Keep Going
Building something from nothing is exhausting. There are days when nothing works, when every line of code seems to introduce new problems, when the gap between where I am and where I want to be feels impossibly vast. I've questioned this path more times than I can count.
But then I'll see someone use Kaer to create something they couldn't have created otherwise. A small business owner who finally has a website that represents who they actually are. An artist whose portfolio site captures the essence of their work instead of burying it in generic templates. Those moments make everything worth it.
There's also something deeper, harder to articulate. I believe we're at an inflection point in human history. The tools we're building now — truly intelligent, genuinely autonomous AI systems — will reshape what's possible for our species. I want to be part of that. Not as a passive observer, but as someone actively shaping the direction we take.
So that's why I'm building Kaer. It started with frustration, evolved through countless failures, and continues because I genuinely believe this matters. The journey is far from over. But for the first time, I can see where we're heading. And I can't wait to get there.