Today AI interfaces act like digital printers.

They take a prompt and print a static page of text or a fixed layout. Every time your intent shifts, the system has to reprint the entire page, breaking cognitive flow.

ChatGPT gives you walls of text. Google gives you lists of links. TikTok gives you endless scrolling.

None of these work for how people actually explore information.

The industry treats this as solved and focuses on building better models.

We think that's backwards. AI models will eventually become a commodity, like water or electricity. The interface layer, not the model, will determine who wins.

We are backing a cracked technical team going after one of the biggest markets in computing, thinking from first principles to build Generative UX, the next major paradigm shift in how humans interact with the AI-enabled web.

Heywa Labs has the team to do it.

Building GenUX

LLMs can do so much more than static outputs. They understand intent, track context, generate structure on the fly. You could build interfaces that reshape themselves around how you're thinking instead of dumping fixed formats at you.

But while the models have gotten smarter the interfaces have stayed the same.

Interfaces have much more potential than just visual design. Systems can reshape themselves around how you're thinking instead of just dumping information at you in a fixed format.

Heywa is moving toward this, and it’s called Generative UX (GenUX).

How Heywa Labs is Different

Most attempts at adaptive interfaces fail because they try to generate everything from scratch every time, resulting in slow, unpredictable, broken experiences.

Heywa Labs built a fixed set of components - cards, comparisons, timelines - that the system arranges and rearranges instantly. These predefined building blocks then reconfigure on the fly, which means their interfaces load in two seconds while other "adaptive" systems take ten.

The components aren't just visual either. Each one carries meaning the system understands - this is a comparison, this is a deep dive, this invites exploration. The interface doesn't just look different, but behaves differently based on what you're doing.

The products will be used for various applications, over time, for everyone.

The Product Vision

Heywa Labs is the company.

Heywa is their first product - a consumer search app that feels like ChatGPT meets Pinterest. Visual, adaptive, and mobile-first.

The app generates structured experiences made of tappable cards that evolve instantly as you explore. Each interaction deepens the system's reasoning or reveals more context without wiping the screen. This works better for how people actually explore topics like cooking, travel planning, DIY, fitness - activities not well served by paragraphs of text or lists of links.

But more than a product Heywa is a proof point for what GenUX can be and a laboratory for refining how adaptive interfaces should feel at scale. The GenUX system powering it is designed to generate adaptive interfaces across different contexts far beyond search.

This is the foundation for many products. Heywa search is the first manifestation.

Search as the Wedge

Search is the perfect proving ground for this vision. The market is massive, the incumbents are vulnerable, and the interface problems are glaring.

Search has long been an uncontested category. Google became the verb, and at its peak, commanded 94% market share, operating with unprecedented category dominance. But the castle is eroding. Google's global search market share dropped last year to 89.57%, the largest single-year drop in a decade.

Search with AI has moved from information retrieval, sifting through dozens of blue links, to knowledge synthesis, getting direct answers. Since 2022, Generative AI has dramatically affected the search economy. AI discovery has reduced the need to click through to websites, leading to a 60% decrease in click-through rates, slashing organic traffic for major publishers up to 40%.

But the fragmentation runs deeper. YouTube “how-to” content serves over 500 million views daily and social platforms like TikTok and Instagram are increasingly outpacing Google search, with over half of Gen Z now using them as their primary search starting point.

The Audacity of Taking on Search

When Milena Nikolic and Martin Sturla pitched us, Perplexity already had 30 million users. But they weren't daunted by the space because they are all in on building a new search experience, one that's only possible with LLMs but fundamentally different, focusing on interaction design, not model capability.

This audacity is exactly what we look for at Cherry, particularly when backed by elite technical depth and execution speed.

We funded them in May and seven months later they had a working LLM search product live in the App Store in the US, UK, and Canada. That execution velocity on a problem this complex is extraordinary, shipping faster than teams ten times their size because they've built GenUX from first principles as a system that can generate, adapt, and maintain interactive experiences dynamically.

What's even more audacious is that search is just the opening move. The GenUX system they've built can power entirely new interfaces beyond search. This is their wedge into building the interface layer that could underpin millions of experiences across the web.

Heywa Labs represents one of Europe's clearest attempts to define a new AI interaction paradigm. While most of the industry races to build better models or apply existing ones to vertical use cases, Heywa Labs is rethinking the fundamental interface layer itself. If models become commoditided infrastructure, whoever controls the interface layer controls how billions of people experience AI. This is the ambitious technical bet Europe rarely gets credit for making - and Heywa Labs is proving that the best AI products come from solving hard interface problems, not just training bigger models.