A multichain wallet from concept to 2,000+ beta users, as the only designer

EXTRA WALLET · Non-custodial multichain wallet with a dApp store and an offline mode

Context

A non-custodial multichain wallet with a dApp store and an offline mode: you sign a transaction on a device with no internet, so your key never touches the network. I was the only designer, working with 8 developers.

Problem

The product packed many features into one wallet, and people got lost in them. The hardest part to understand was the offline mode.

Company

EXTRA WALLET

My role

Lead Product Designer (part-time)

Team

Sole designer, 8 developers

Platform

Web app + offline app, 5 breakpoints, light and dark themes

Timeline

Nov 2024 – Apr 2026

2,000+

users in the closed beta in its first months (company data)

500+

screens designed in 13 months, from concept to beta

50%+

faster design iterations with an AI-assisted workflow

Reworking a flow with the whole team

I ran 5 workshops with the team to unblock stalled decisions. In one, we reworked the HD wallet scan, which finds a wallet's addresses and balances from a QR code.

I asked everyone to walk the flow as a first-time user and note where they'd get lost. From their notes I built a shorter flow: registration moved out of the way, the scan is explained before it starts, and advanced settings became optional.

Testing the wallet with real people

In 5 usability sessions, people thought they were signing up for a service, while the product expected them to import a cryptographic key. They met four new terms at once, didn't know what a seed phrase was, and couldn't find Swap or WalletConnect.

I split Create and Import, explained the seed phrase in one line, moved offline import out of the main path and rewrote staking labels in plain words. All of it shipped in the beta.

A design system for 500+ screens

I built the design system from scratch: 30+ components with size and state variants in both themes, across 5 breakpoints.

Designing faster with AI

I set up an AI-assisted workflow with Claude, Cursor and a custom Figma skill for component updates and responsive adaptation. It cut my design iteration time by 50%+.

Reflection

What I'd do differently

For the beta, I'd focus on 2–3 core features and design each one end to end, with onboarding and clear explanations. Starting small serves people better than 10+ features that aren't polished yet.