Blink

Blink reimagines the translation and language learning experience. Instead of translating entire sentences as large blocks, Blink comprehends language more naturally by intelligently extracting meaningful contextual blocks and emphasizing what matters most. By combining contextual translation with context-based learning, Blink reduces cognitive load and transforms everyday conversations into seamless, meaningful learning moments.

Problem to solve

Translation creates cognitive overload

Traditional translation tools treat language as complete sentences, often presenting users with large blocks of translated text. While this provides an answer, it places a significant cognitive load on users, especially during live conversations, making it difficult to quickly identify and process what matters most.

Translation and learning are disconnected

Many translation users are also language learners, yet translation and language learning are often treated as separate experiences. Users must deliberately switch between the two, creating friction and missing opportunities to learn naturally through everyday conversations.

Design Challenge

How might we make translation more intuitive and closer to how humans naturally interpret language, so users can focus on what matters, understand more easily, and seamlessly turn translation into a learning experience?

Design • Engineering • Psycholinguistics

What matters most?

What matters most in understanding language, is it key words, proficiency levels or semantic? How to extract them?

Research indicates that people comprehend language by integrating words within their broader semantic and contextual relationships, rather than processing each word in isolation. While keywords identify individual words and proficiency levels indicate potential learning difficulty, neither captures what is most meaningful for understanding a specific sentence or conversation. Semantic blocks allow Blink to reflect how meaning is naturally constructed, helping users focus on what matters while creating more relevant opportunities for learning.

Coded Prototype

By leveraging Cursor, design and development evolve together through rapid, continuous iteration, reducing the compromises in traditional workflows. This integrated approach enables faster experimentation, more seamless refinement, and a more cohesive end-to-end user experience.

Streaming semantic pills

Voice recognition transforms speech into text, then intelligently breaks it into meaningful semantic pills. These pills appear progressively in real time, keeping pace with natural speech to create a seamless, low-latency experience that reduces cognitive load and makes spoken information easier to follow and process.

Weighted and highlighted pills

Each semantic pill is evaluated through a contextual weighting system that assesses its meaning and relative importance within the sentence, helping users identify key information at a glance and reducing cognitive effort.

Next steps

Next steps focus on refining the semantic slicing and weighting systems to deliver a more intuitive and context-aware experience. An “Add to Favourites” feature will also be introduced to support long-term learning, allowing users to save, revisit, and reinforce knowledge retention over time.