Natural queries
Let people begin with the words they already have instead of a form to configure.
01 / iOS utility / Product design & engineering
UnitX
Making everyday conversion feel immediate
A unit converter designed around the way questions actually appear: type it, say it, scan it, save it, or pick up a familiar conversion from your wrist.
Context
Conversion is usually a small task that happens at an inconvenient moment.
A traveller checks a price, a cook adapts a recipe, a shopper reads a label, or a student sees an unfamiliar measurement. Most converter apps make people choose a category, pick two units, enter a value, and only then see an answer. UnitX was designed to shorten that setup without losing control when accuracy matters.
- Audience People converting measurements or currencies in the middle of another task
- Product challenge Make the fast path feel immediate while keeping exact manual conversion discoverable
- Platforms iPhone with an Apple Watch companion for repeated utility moments
Design choices
Start with intent. Offer control when it helps.
The primary surface accepts short natural requests such as “72 kg to lb” or “5 miles in km.” Manual controls remain a parallel route rather than a buried fallback, while voice and camera entry remove typing when the number is spoken or printed.
Manual precision
Keep categories, units, swapping, and values explicit for exact or repeatable work.
Voice and camera
Turn a spoken request or visible label into an editable starting point.
Evidence
A single converter, with several useful ways in.
The product keeps one clear outcome—the conversion answer—while adapting its entry method to the moment. The app screen shows intent-led suggestions and access to manual, microphone, and camera flows; the iPad view makes the navigation model visible at a larger scale.
Reflection
Flexibility works only when the default remains calm.
UnitX deliberately makes the common request feel light while retaining reliable escape hatches for manual input, editability, history, favourites, and Apple Watch follow-through. The next iteration should validate recognition recovery and which repeat conversions are genuinely worth surfacing on the wrist.