Dyneti / designing a card-verification flow that feels instant.
Dyneti uses deep learning to verify payment cards in seconds. The design problem is making a fast, opaque ML decision feel fast and trustworthy at the same time, on a mobile camera surface where the user has no patience for friction.
Designing a card-verification flow that feels instant on a surface where users have zero patience.
Dyneti uses deep-learning models to verify credit cards in real time and prevent fraud at the point of capture. The design problem: make a fast, opaque ML decision feel fast and trustworthy on a mobile camera surface where the user has no time to read instructions.
We partnered with Dyneti to design DyScan Protect, a secure verification flow that works across both mobile and desktop. The constraint shaped the entire interaction model. The flow needed to guide users step by step, recover gracefully from camera errors, and clear in under ten seconds, while Dyneti's model analyzed the card behind the scenes.
What we shipped.
- 1 week from kickoff to functional prototype.
- Under 10 seconds average verification time per card scan.
- 4 verification flows designed and validated, covering single-sided, two-sided, mobile-only, and QR-handoff variants.
Cross-device verification without the friction of "open this on your phone."
Many users start a transaction on desktop but have their card and camera on their phone. Forcing them to switch devices manually breaks the flow. We designed a QR-based handoff that turns the device switch into a single scan.
After acknowledging the security prompt on desktop, the user scans a QR code displayed on screen. That opens Dyneti's mobile scanning interface directly on their phone, no app install, no account linking. The card scan happens on mobile with the phone's camera, and the verification result completes the desktop transaction.
The cross-device flow keeps the verification process simple, even when the user starts somewhere the verification cannot happen.
What happens when the lighting is wrong, the angle is off, or the card is partially obscured.
The success path is the easy design problem, the error path is where verification flows live or die. Users blame themselves for camera failures, then give up. We designed the error states to do the opposite: tell the user exactly what went wrong, show them how to fix it, and let them retry without restarting the verification.
- Visual feedback that names the failure cause: lighting, positioning, camera focus.
- Simple, single-action retry, no return to the start.
- Inline guidance for repositioning, with the camera view kept visible the entire time.
- Strict verification requirements maintained, no compromise on fraud detection for the sake of convenience.
Fraud detection succeeds or fails on the recovery path, not the happy path. The happy path is what the user sees once. The recovery path is what they remember.