Circle · 2025–26 · Lead Product Designer

Arc Portal

An Arc-native financial account: fund, earn, swap, send, and give an agent a wallet.

Consumer fintechOnchainAgent walletsPrototype to productionLive product ↗

Hours
From product definition to a working prototype in the repo: 11
User interviews synthesized into the launch iteration: Mainnet
Launched with Arc

The problem

Arc needed a front door for people, not just developers. Someone arriving at a new chain has the same three questions every time: how do I get money onto it, what can I do with it, and how do I get it back out. Then a fourth, new question for 2026: how do I let my agent use it.

Portal is the answer to all four in one place.

My role

I defined the product, built the first working prototype, ran the research, and iterated the launch version. After launch I took over primary design coordination for Portal.

Defining the product

The first job was deciding what Portal was. Not a wallet, not an explorer, not a DeFi dashboard, but an Arc-native financial account: one surface spanning wallet funding, earn, swap, send, ecosystem discovery, and agent wallets.

That framing did a lot of work. It kept the team from building six apps, and it gave every feature a shared mental model: this is your account on Arc, and these are the things an account can do.

Portal's home at mainnet. The navigation is the product definition: Earn, Swap, Send, Agent, Discover.

Prototype first

With the definition agreed, I built a working prototype directly in the repository, in a few hours rather than a sprint. It wasn’t polished, but it was real: real navigation, real state, real flows you could click through on a phone.

That prototype became the artifact for everything that followed. Leadership aligned on it. Engineering scoped from it. And it’s what we put in front of users.

Research

I ran and synthesized 11 user interviews on the prototype and early builds. The signal was clear on both sides:

What worked: visual trust, simplicity, the swap flow, and the agent experience. People understood “give your agent a wallet” faster than we expected.

What didn’t: the difference between a wallet balance and an Arc balance, what happens cross-chain, how risky “earn” actually is, and how withdrawal works. All four are conceptual, not visual; all four fed the launch iteration and the post-launch queue.

Earn: third-party vaults with Circle Guarded labeling, deposits shown before APY so the risk framing comes first.

Swap: the flow research liked most. Balance, slippage, and the liquidity source are all visible before 'Get started'.

Agent: create a wallet with a spending cap, hand the setup prompt to Claude Code or Codex, and it transacts on its own. Example prompts show what 'on its own' means.

Discover: the ecosystem, so Portal is a front door to Arc rather than a walled garden.

Agent wallets

The agent flows were the part of Portal with no precedent. I designed the setup and funding experience around what a person actually needs from an agent’s wallet: let the agent spend, see what it’s spending, and cap it. Setup, monitoring, and limits are one flow, not three settings pages.

Outcome

Portal launched as part of Arc mainnet and now exposes earn, send, swap, discovery, and agent-wallet functionality. Post-launch, the research findings became the iteration roadmap, starting with the balance model and withdrawal clarity.