Today, the main barrier isn't core technology - it's the lack of smart, simple, and intuitive interfaces that truly empower people to use it. We design the interface first, then put the model behind it. Chat is one answer. We're interested in the others.
We design, prototype to learn what works, then build. We work at the intersection of AI capabilities and the interfaces people actually use - from search and discovery to structured creative tools.
People stopped searching Google and started talking to LLMs - but open an online store and the main tools are still dropdown filters or, at best, a chat box grafted onto the same old layout. We design interfaces that let people find things the way they actually think about them: by mood, by reference, by example.
Typing into an empty chat window is a poor way to do most things. For shopping, designing, or planning, people need to see options, compare them, and adjust by feel. We build interfaces where parameters are visible and controllable - not hidden behind a blinking cursor.
Shopping interfaces haven't kept up. Most still run on dropdown filters and keyword search. Newer ones just bolt a prompt box onto the same layout - treating AI as a feature, not a reason to rethink the experience. We're building what comes next - starting with fashion as our first proving ground.
Shoppers describe what they want in loose terms - "something for a wedding in Tuscany," "cozy but not frumpy." Filters throw all of that away, and prompt boxes turn it into a guessing game. We turn open descriptions like these into signals the catalog can actually match against.
Most shopping isn't keyword-driven, it's visual. People find something close and want "this, but longer, darker, looser." We build interfaces that take images and examples as the starting point - not as an afterthought pasted into a chat.
Product tags are usually written to rank on Google, not to describe what something actually is. We restructure catalogs around attributes a model - and a person - can reason over: style, formality, mood, cultural reference.
Mood-based visual browse over manufacturer and marketplace catalogs. Research, methodology, and live demos.
To support more natural, visual, and exploratory user–system interaction, we work across the full stack from data architecture to interface design.
Re-designing how information is organised - so it can support AI-native interaction beyond classic filters, dropdowns, and keyword queries.
Indexing content with vision and language models - so it can power semantic search, visual matching, and recommendations instead of just keyword lookup.
Enriching datasets with visual and contextual content that can be reused across interfaces, search, and discovery.
Designing interfaces that make AI capabilities visible and steerable - so users can browse, compare, and adjust by feel instead of guessing the right prompt.
A small team with wide curiosity. We build things, learn from them, and work on a few projects at a time.
Want to know more? Get in touch.
We are currently working on solutions for e-commerce, marketplaces, and online shopping. If you run such a business and would like to learn more, we'd be happy to talk. We also welcome research partnerships and investor conversations.
