All workSimeon Daniel
Web3

Tidewallet: a crypto wallet that explains itself

People rarely lose money in crypto to a hack. They lose it in a confusing moment: a wrong address, a fee they didn’t expect, a token they didn’t understand. Tidewallet puts AI at exactly those moments, and only asks for money after the free checks have proven their worth.

My role
Product design, end to end
Platform
iOS mobile app
Focus
Activation, swap completion, Pro trial starts
Tools
Figma

01The business problem

Crypto wallets lose people at the moments that feel risky: setting up, pasting an address, and confirming a swap full of fees nobody explains. A wallet people don’t trust gets used once and then emptied.

For a wallet business that earns from swap fees and subscriptions, every abandoned send or swap is lost revenue, and every scam that gets through is a lost customer.

02The metric

One number to move, and one I refuse to hurt.

PrimaryPro trial starts after the free safety checks run out
SupportingFirst funded send in week one, and swap completion
GuardrailSend completion must not drop because warnings add friction

03What people actually do

Under uncertainty, people do one of two things: freeze, or click through without reading. Neither is safe.

Generic warnings stop working fast. The body habituates to repeated alarms the same way it stops noticing a ticking clock, so a wallet that warns about everything trains people to ignore the one warning that matters.

A specific reason (this address is two days old) is believed. A vague one (proceed with caution) is skipped.

04Hypotheses

Untested. Each one is written so it can be proven wrong.

H1

Passkeys instead of a seed phrase

More new users fund their wallet in week one, because the scariest, most permanent-feeling step disappears.

H2

A reasoned safety check with a $5 test send

Fewer abandoned sends and fewer sends to the wrong person, because a small, reversible step lowers the cost of being wrong.

H3

Fees explained before the button

Higher swap completion, because surprise is what triggers the exit.

H4

Pro offered after the checks catch something

More trial starts without hurting retention, because people pay to keep a benefit they’ve already felt.

05Design decisions

Activate

Sign-up without a seed phrase

What I did
Passkey and face sign-in are the main path, with the white button doing one thing: create the wallet.
What I chose not to do
Asking for a 12-word seed phrase on day one. Backup still exists, but it moves to later, once there’s money worth protecting.
Retain

A reason to open the app on a quiet day

What I did
One useful AI tip sits at the top of the history sheet, written in plain language about the person’s own activity.
What I chose not to do
Sending tips as push notifications. Non-urgent alerts teach people to ignore alerts, and the safety warnings depend on being noticed.
Convert, safely

Evidence, not accusation

What I did
The AI shows three specific findings on a dashed line, so the warning reads like evidence. The safe option is the white button; sending everything stays available, just quieter.
What I chose not to do
Blocking risky sends outright. People would move to a wallet that lets them, and the decision should stay theirs.
Convert

No surprise fees

What I did
Every fee is named, priced, and explained before the Review button: network fee, swap fee, and possible price movement.
What I chose not to do
A single total cost line. It looks simpler, but it hides which part of the cost is avoidable.
Monetize

Ask after value, not before

What I did
The paywall opens with what the free checks already did, keeps “Not now” visible, and preselects yearly because it’s the better deal for someone already getting value.
What I chose not to do
A paywall at sign-up. At that point the person is paying for a promise, not for something they’ve seen work.

06How AI is used

AI explains what’s about to happen and suggests a safer option, but it never acts on its own. Every send, swap, and signature stays a human decision. It names the risk, shows its reasoning, and offers one next step.

07How I’d test it

Events I’d track from day one

  • wallet_created
  • first_funded_send
  • safety_check_shown
  • test_send_chosen
  • send_abandoned
  • swap_fees_viewed
  • swap_confirmed
  • paywall_viewed
  • trial_started
  • A/B test the safety check: specific reasons versus a generic warning, measuring abandoned sends and wrong-address reports.
  • A/B test paywall timing: after the free checks have caught something versus on day one, measuring trial starts and 30-day retention.
What success looks like. Trial starts rise in the after-value group while send completion holds steady.

08What I’d do next

  • Let experienced users turn off the $5 test suggestion after a run of safe sends, so it never feels patronising.
  • Design plain-language explanations for signing smart contracts, the riskiest moment in Web3.