Work

Web

iOS

·

2026

Icovela

Creating a custom icon often means choosing between generic libraries, inconsistent AI outputs, or time-consuming manual design. I designed and built Icovela, an AI-powered platform that helps creators and product teams generate consistent, ready-to-use icons from simple text prompts.

Role

Product Design & Development

Credits

Solo Project

Next project

Code Blue Sim

Problem

More than 36 million developers joined GitHub in 2025, and 84% of surveyed developers use or plan to use AI tools. Yet inconsistent output remains developers' biggest AI frustration. As building gets faster, creating polished, cohesive visual assets remains a bottleneck.

Outcome

Raised user activation from 24% to 63%

From product strategy and UX/UI to AI generation, subscriptions, and analytics, I independently built Icovela into a production-ready product. Funnel-driven iterations later increased registered-user activation by 2.6×.

Product Overview

How I got there?

Design Process

Opportunity

I began with user interviews and competitive research to understand how creators and independent developers source icons for their products. Existing icon libraries were fast but offered limited customization. General-purpose AI tools offered more freedom, but required users to translate visual ideas into complex prompts. This revealed an opportunity for a more focused, approachable icon-generation experience.

How Might We

make custom icons fast, distinctive, and easy to create?

Define the Product

1. Make Generation Approachable

Users should be able to begin with a simple idea rather than learning how to write complex prompts.

2. Make Style Selection Clear

Visual examples should help users choose a direction before committing to a generation.

3. Minimize Time to First Result

The path from landing on the product to generating an icon should feel immediate and easy to understand.

After launch

Test & Feedback

Creating a Feedback Loop

To learn from real usage, I instrumented the journey from first visit to download and brought the signals into an admin dashboard. This allowed me to see not only how many people signed up, but whether they reached the product's first meaningful action.

Visitor > Sign up > First generation > Download


Turning Drop-Off into Activation

After launch, only 24% of registered users reached their first generation. The funnel showed where users stopped; reviewing the experience and user feedback pointed to two sources of friction.

01 Make Free Value Hard to Miss
Before

The offer was present, but visually secondary.

After

A consistent value message guides users from discovery to sign-up.

02 Make the Fastest Path the Default
Before

Custom Style was the default, so reference images were required before anyone could generate.

After

Isometric 3D is the default, so a prompt is all it takes. Custom Style stayed for anyone who wants more control.

Outcome

Activation Increased 2.6×

Sign-up to first generation

24% to 63%