AI Task Manager. Turning ‘write a paper on coastal cities’ into five small wins.
AI Task Manager is a mobile-first concept that explores how generative AI can sit inside a personal todo app — not as a chat sidebar, but as a single, well-placed action that turns vague goals into clear, editable steps.
A friendly, single-column intake — title, priority, tags, optional description and a clear ‘Break into Steps’ moment that hands the heavy lifting to the AI.
An honest loading state — the AI says exactly what it’s doing, so the wait feels intentional rather than broken.
AI-generated subtasks the user can edit, reorder, delete or add to — the model proposes, the human disposes.
Priority, due date, tags, description, sub-tasks, reminder and an overflow for duplicate / share / delete — everything you need on one calm screen.
A small ‘Great Job! 🎉’ toast and a celebratory bar — completion gets a moment, not just a checkmark.
The context.
- 01
Writing a task is easy. Knowing how to start it is the actual problem — most users stall on goals like ‘research a topic’ because there’s no obvious first step.
- 02
Most AI productivity tools live in a separate chat tab — a context-switch the user has to opt in to. The model needs to live inside the flow, not outside it.
What we set out to achieve.
AI in the flow, not in a sidebar
‘Break into Steps’ sits exactly where the user finishes describing the task — one tap turns a goal into a draft plan.
Human stays in control
Every AI-suggested step is editable, reorderable and deletable. Users can also append their own. The model proposes, the human disposes.
Honest loading, soft delight
‘Thinking through this…’ tells users exactly what’s happening; ‘Great Job 🎉’ rewards completion. Tone-of-voice is the smallest, biggest part of the system.
Calm, scannable density
Soft greens and warm neutrals; rounded chips for priority, date, tags; a single column that never feels cramped on a one-handed grip.
How I worked through it.
Mapped the ‘blank page’ moment
Identified the highest-friction beat in any productivity app: the user knows the goal but doesn’t know the first step. That moment became the product’s core opportunity.
Designed a single-column intake
Title, priority, tags, optional description — paired with a clearly labelled ‘Break into Steps’ button.
Made the AI moment legible
An explicit ‘Thinking through this…’ loading state with a calm spinner. Then an editable list of suggested steps, with drag-handles, numbers, delete-X and an ‘Add Another Step’ field at the bottom.
Designed the full task lifecycle
Task details with priority, due date, tags, description, sub-tasks, reminder toggle and an overflow menu (duplicate · share · delete). Completion is celebrated with a strikethrough title, a green progress bar and a ‘Great Job 🎉’ toast.
The outcome.
A concept flow where AI removes the scariest part of starting (‘what do I even do first?’) without taking away control, ownership or the small rituals — celebration toasts, reminders, soft-coloured tags — that make a task app feel personal.
What I’d carry into the next one.
AI is most useful when it removes the blank page — not when it tries to remove the user’s judgment.
Tone-of-voice (‘Thinking through this…’, ‘Great Job 🎉’) is the cheapest, most underrated way to make an AI feature feel trustworthy.
Soft greens, rounded chips and a single-column rhythm make a task app feel personal.