Focus areas
From models to products, from architecture to iteration: four areas where we put our engineering and design effort, and where you can see them at work.
AI Application Development
Turning model capabilities into reliable product experiences across reasoning, retrieval, and personalization.
A capable model is not yet a product. We shape the task, ground answers in the right context and design the interface around the result, so the output is something people can rely on.
- Understanding messages, memes and screenshots in context, as MemeWingman does
- Generating lesson materials, practice and flashcards from a syllabus or a learner’s own notes, as StudyOS does
- Scoring spoken answers and explaining what to improve, as 入戏英语 does
Intelligent Learning Experiences
Interaction, content, and feedback designed around real learning paths — making study more visible and effective.
Good learning tools make progress visible. We design interaction, content and feedback around how people really practice: in context, with active recall, and with feedback that explains.
- Learning in context: practicing spoken English inside real TV scenes
- Active recall and spaced repetition with smart flashcards
- Feedback that names the one thing most worth fixing
Seen in
Global Product Architecture
Scalable architecture and engineering practice for multi-language, multi-region, and multi-payment environments.
Products for global users need more than translated buttons. We plan for multiple languages, regions and payment environments from the start, so a product can grow into new markets without a rewrite.
- Multilingual experiences: MemeWingman explains in five languages while meme names stay in their original language
- Region-aware products, designed for users across the US and East Asia
- App and device ecosystems: Boilon manages AI audio devices from a mobile app
Seen in
Data-informed Product Iteration
Refining critical user paths and core experiences continuously, grounded in metrics and real feedback.
We choose a small number of key user paths, measure them and keep refining them with real feedback. Metrics tell us where to look; users tell us why.
- Define the core path of each product and watch it closely
- Pair quantitative metrics with qualitative feedback
- Ship small changes often and keep what works
Seen in
Have a partnership, a product question or an idea?
Send us a note and we will get back to you by email.