Web Application
Dhad — Arabic Typing & Grammar Tool
A browser-based tool for Arabic writers — grammar checking, diacritics auto-insertion, and style suggestions powered by NLP.
- Year: 2023
- Duration: 4 months
- Client: Independent Media Publication
- Outcome: 40k monthly active users
Technologies: Next.js, Python, NLP, Arabic Processing
The Challenge
Arabic writers working on digital content had no native browser tool for grammar and style checking. Existing tools were either English-only or required desktop software installs. The client wanted something as frictionless as Grammarly — but built specifically for Modern Standard Arabic.
Our Approach
- Corpus & NLP Research — Partnered with an Arabic NLP researcher to build training data from 50k verified Arabic articles. The grammar model was trained and evaluated before any UI was designed.
- Editor Experience Design — The writing editor was the entire product — we spent 6 weeks on it. Tested three interaction models for inline corrections before settling on the approach that felt invisible.
- Performance & Scale — The NLP pipeline needed to return suggestions in under 400ms. Optimised Python inference, added Redis caching, and load-tested to 5k concurrent users before launch.
Core Features
- Real-Time Grammar Check — Underlines grammar errors as you type with one-click corrections. Supports MSA and formal Arabic conventions.
- Auto-Diacritisation — Intelligently adds short vowels (tashkeel) to selected text using a context-aware model trained on classical and modern Arabic.
- Style Suggestions — Flags passive-voice overuse, sentence complexity, and word repetition with editorial-grade suggestions.
- Focus Mode — A distraction-free fullscreen editor with word-count goals, writing streaks, and a dark/light toggle.