The problem

The widgets available to you usually depend on what your installed apps provide. Weather, calendars and battery levels are easy to find; a reminder that fits a particular routine may take more work. A plant-watering countdown, a reading tracker or a note about dinner can be simple ideas, yet still require a template editor or a separate app. This project explores whether describing those small needs could be a more approachable way to personalise a Home Screen.

The insight

Here is the constraint that shaped everything. An app cannot place a widget on your Home Screen. iOS does not allow it. The user has to leave, press and hold the wallpaper, find Edit, open the gallery, and pick the thing they just made. So the most important moment in this product happens in a place I do not control, in an interface I did not design, after the user has already closed the app. That inverted the job. The app is not the destination, it is a workshop. Every screen exists to get someone to a finished object and then hand them off cleanly, without losing them at the door.

The idea

Many widgets share the same styles and purposes because they come from a fixed set of app templates. I wanted to explore a different approach: describe what would be useful to you, then shape the result around your own routine. The concept makes that process a conversation. Start with an idea, review the preview and ask for changes until the widget feels useful and personal.

The architecture

The whole product is one path: open, describe, refine, save, hand off. My Widgets is the library. New Widget is the workshop. Widget Details is where a finished object lives. Profile holds the account and the credit balance. There is no browsing, no template gallery, no community feed. A template gallery would have been the easy version of this product and it would have rebuilt the exact problem I started from: choosing from what somebody else made.

Design system

The visual design draws on Apple’s iOS design resources, using familiar buttons, toolbars, keyboards, sheets and pickers. A widget app should feel at home in the system where its widgets will live. Working within an existing design language shifts the focus to spacing, hierarchy and sequence. The interface supports the preview and the prompt, so the object being made stays at the centre of the experience.

Flow walkthrough

Create with a prompt, review the result and add it to your Home Screen. The onboarding introduces that simple sequence before sign-in and email verification, so the purpose of the app is clear before an account is needed.

Small, medium and large are not a setting, they are the first decision. A countdown that works in a small square is a different piece of writing from one that fills a large one. Asking for the size first makes the preview honest from the first keystroke, and the placeholder in each screen is already at the true widget ratio.

The wireframes separated Improve it and Polish it into two steps. In the final flow, I brought them into one persistent field: Want to change anything? A status label shows the transition from Generating to Ready. Keeping that field available makes refinement an ongoing conversation rather than a fixed number of steps.

Widget Details ends with Add to Home Screen, which does not add the widget to the Home Screen. It cannot. So instead it does the only honest thing: three numbered steps, in the user's own words, for the thirty seconds they are about to spend outside the app. This was the hardest screen and it is the one I would defend first. Most products would hide this behind a cheerful animation and let people fail silently in the widget gallery.

Couldn't create this widget. Check your connection and try again. The prompt stays in the field. The error sits inline, above the input, not in an alert that has to be dismissed before you can see what you typed. Retry is one tap and does not make you write it again.

The settings concept shows the remaining AI credits alongside appearance, default widget size and haptics. The aim is to make the generation allowance visible before someone starts creating. The displayed balance is an example; the credit system and pricing would need validation.

Close

This is a concept. The main assumption still needs testing: can a short description become a useful widget, and can people successfully add it to their Home Screen?

1. Can people refine an imperfect result? Real prompts may be vague or combine several requests. I would test whether the preview and refinement field help people reach what they meant.

2. Do people finish the handoff? The numbered instructions are a design proposal. I would measure whether people can follow them and find their widget.

3. Is a credit balance understandable? The example balance is not a pricing model. I would test how people understand generation costs before setting limits.

What I learned

Designing inside Apple's system is more constrained than designing a brand, and better practice. You cannot solve a hierarchy problem by picking a nicer typeface. The only tools left are order, spacing and what you decide to leave out, which turns out to be most of the job.