✦ My Role

I joined Coco at the very beginning, when it was still an idea on paper. Working alongside the team, I helped define the flows and structure, and designed the screens and the brand around them.

Six years later, I came back to redesign the product. This time, the starting point was different: the core method already existed and didn't need to be reinvented. My job was to figure out why the product wasn't fully working, rebuild the system behind it, how users moved through each day, and what the product was actually for, and then design every screen around that.

✦ The Challenge

Coco's method already worked. What was missing was a clear line for users to follow.

Opening the app meant landing straight inside a library of songs, with nothing telling you what to do next or where any of it was going.

• In testing, roughly 8 out of 10 people never got past the very first move: open a song, tap a word you don't know. That's as far as most sessions ever went.

• Nothing ever marked a day as done. Without that small close, there was nothing pulling anyone back tomorrow.

• There was no visible sense of progress toward anything. Getting better at Coco had no shape, so there was nothing to stay motivated for.

None of this was a missing tutorial. Coco had a method that worked, and no way for users to see where they stood inside it, or why to come back.

✦ My Approach

Our first instinct was to add structure, give Coco a clear path to follow. But turning it into a course with fixed lessons would solve the confusion by killing what made it worth having: the freedom to learn from what you already like, not from whatever someone else decided you should study. Forcing that into a fixed path would only turn Coco into just another app in a crowded market.

The learning method we already had proved itself. Real content, reviewed, put to use. The hypothesis behind it held up too: you could learn a language from the whole internet. The problem wasn't the freedom we gave users, it was that the freedom never explained itself. Coco wasn't a course, but it didn't say what it was, either. The question stopped being how to structure Coco and became something else: if it's not a language-learning app, what is it?

The answer was a workspace: a place to organize and manage the language you're already learning. That shift changed how we talked about the product, who we were building it for, and what we built next.

With that clear, giving Coco a daily routine stopped feeling like the same risk as before. It was no longer about boxing users into someone else's course. It was about giving shape to their own workspace. Every morning, AI builds a short plan around whatever the user already chose to watch, listen to, or read:

• Learn a handful of new words from that content.

• Put them to use and get feedback on your mistakes.

• Review everything with spaced repetition.

A system of modals hands you from one step to the next, and a visible streak turns every finished day into a reason to come back tomorrow.

We also built a Chrome extension that carries the same workspace idea outside the app. Click any word, on any site, and it lands straight in the same flashcards you review inside Coco. The more of the internet someone already lived in, the more they had to learn from, and the more they had to organize inside Coco.

Coco didn't need to teach English. It needed to help you organize what the world was already teaching you.