Imagine you want to push yourself — and bring a few friends along.
What if starting that challenge took just one sentence?
When I joined the team, the previous product designer had already left. I inherited an existing product and its design work in a startup where the team had prioritized rapid shipping over scalability. The experience had accumulated inconsistencies and friction as features were added.
Napoz is a social mobile app where anyone can start a group challenge — from a 30-day streak meditation to a friendly burpee contest or a neighbourhood café hunt — invite others to join, and share progress together.
The founder brought an urgent business concern:
“Weekly active users declined 10% in the past month. We need to increase the user engagement rate.”
As the sole product designer, I took ownership of investigating the experience and defining the redesign. To understand where design could support engagement, we looked at the journey that gives people reasons to participate and return.
Creation supplies the activities that the rest of this loop depends on. That made it an important opportunity to investigate: reducing the effort to publish could give people more opportunities to join and contribute.
The design question became:
How could we help more people turn an idea into a challenge others could join?
We reviewed product analytics screen by screen, and Create Challenge stood out. The existing flow required 10 steps and an average of five minutes to publish a challenge.
I conducted user research, and user interviews reinforced three sources of friction:
When flexibility became friction. The problem was how much of that flexibility creators had to configure for themselves.
| User difficulty | Design response | Intended benefit |
|---|---|---|
| Deciding how the challenge should work | Three challenge types with relevant settings | Fewer unnecessary decisions |
| Explaining the challenge to others | AI-assisted drafting with a review step | Less effort to turn an idea into publishable content |
This gave the redesign a clear relationship to the engagement goal: make creation easier to finish so more ideas could become shared activities.
The old flow exposed settings for leaderboards, points, scoring rules, and other mechanics through the same setup process. I stepped back from the screens and identified three recurring challenge models:
The product could handle different challenge structures while helping creators make decisions relevant to their own idea.
The redesign asks what kind of challenge it is, and the screen answers with only the settings that kind needs.

Simpler setup still left creators with the writing. I proposed AI-assisted drafting, but the founder raised concerns about cost and engineering effort. Working with engineering, I built a lightweight text-generation prototype that gave us something concrete to evaluate. It helped secure approval and ship the feature.
The flow settled into three steps: describe an idea, review the generated draft, publish. Creators get a head start without giving up control. I later prototyped the next version of Create with AI — the demo at the top of this page.



As I rebuilt Create, I realized the existing design system couldn't support the product we were designing.
That became its own piece of work — a new brand color, type system and component library — told in Part 02, the Design System.
Part 02 · Design System
What did a simpler Create need underneath it?
A token foundation where every color has one job, a named type ramp, a real button hierarchy, and one Challenge object that keeps its shape on every surface.
Read Part 02
Napoz still supports wildly different challenges. Users just don't have to configure every possibility themselves.
Napoz is live on iOS and Android — see it on the App Store.
I learned to design AI as a reusable interaction pattern:
Challenge creation demonstrated how this approach could reduce manual setup and decision-making. That pattern could become a template for other complex flows.
The broader opportunity is to scale this approach across the product—helping users express what they want while AI handles more of the setup, with user control built in.