One of the most exciting things about Dasher is how it inspires people to build their own variants, ports, and experiments. Over the years we’ve seen implementations in languages from C++ to JavaScript, aimed at purposes from assistive technology to live music composition. Here’s a round-up of notable community projects we know about — if you’ve built something with Dasher, let us know!
Dasher-JS (React) — Jamie Brew
Jamie Brew, creator of the comedy-predictive-text startup Botnik, has built a React implementation of Dasher designed for word-level constrained writing. Where classic Dasher zooms through individual characters, Dasher-JS lets you navigate through whole words — turning the interface into a collaborative writing tool where the language model suggests where your sentence could go next.
Jamie describes the project as emerging from a long-standing interest in predictive text as a “constrained writing game,” and credits Claude Code with making the development pace much faster than before.
Try it: Live demo · Source code
Jasher — a musical Dasher — Jossy Sayir
Jossy Sayir, Associate Teaching Professor at the University of Cambridge and a longtime member of the Dasher research community, created Jasher — a variant of Dasher for composing and generating live music. Instead of navigating through letters and words, you zoom through musical notes and phrases, creating melodies in real time.
Jasher demonstrates how the core Dasher concept — navigating a probabilistic tree by zooming — generalises far beyond text entry.
Try it: Jasher demo · More of Jossy’s Dasher experiments
DasherJava — Jan Schulte
Jan Schulte is re-implementing Dasher in Java as part of his master’s thesis, with a focus on clean, modern, well-documented code rather than a straight port of the C++ original. At around 1,300 lines of code, the implementation includes a working PPM language model, dynamic node creation and deletion in the Dasher tree, and a Java Swing GUI.
Jan has also published a simplified standalone version of the PPM language model for testing and comparison, along with benchmark results showing how the various Dasher language models compare.
The code is not yet fully public while licensing is sorted out, but a snapshot is available via GitFront, and Jan plans to explore an Android port.
Code snapshot: DasherJava on GitFront · Language model comparison · Demo video
Dasher Mobile (Android) — Ján Murin
We previously covered this project in detail. Ján Murin built a native Android implementation of Dasher as his bachelor’s thesis at Masaryk University, with KenLM language model integration and a focus on Slovak language support.
Note: Dasher Mobile is a community project, separate from the official Dasher-Android frontend. Both are built on DasherCore but take different architectural approaches.
Source: Dasher-Mobile on GitHub · Download APK
Build Your Own
All of these projects demonstrate how the Dasher concept — zooming through a probabilistic prediction tree — can be adapted to new platforms, languages, and even entirely different domains like music. If you’re interested in building your own, check out:
- DasherCore on GitHub — the shared engine that powers the official v6 frontends
- Developer handbook — architecture, build guides, and the C API reference
- Feature parity matrix — see what each platform supports today
We’d love to hear what you’re building. Share your projects on GitHub Discussions!