About

In a world where phishing threats evolve faster than traditional warning systems, we set out to redesign the experience from the inside out. Instead of passive alerts that overwhelm or confuse users, we built a frictionless, AI-assisted decision flow that’s fast, clear, and trustworthy.

The result: reduced phishing success, faster decisions, and dramatically lower noise for security teams.

Scop

Platform: Web and Mobile email clients
Timeline: 12-week pilot · 2 UX cycles · cross-functional
Team: Product Design Lead (me) + 2 Designers, 1 Researcher, PM, Engineers, AI Lead, SecOps

The Problem

Legacy phishing warnings are vague, repetitive, and frequently ignored.
Users either dismiss alerts or flood IT with false reports. Security suffers on both sides.

Challenge

Design a clear and trustworthy interaction around AI phishing detection:

• Guides the user confidently, even under pressure.

• Explains the risk transparently without overwhelming.

• Improves over time with user feedback

Solution

A modular, AI-driven interaction flow that appears contextually inside the inbox:

1. Inbox Risk Indicator
Inline badge: “85% likely phishing”

2. Suspicious Email Card
Three clear options:
→ Report
→ Safe Preview (isolated view)
→ Looks Safe

3. Safe Preview
A read-only snapshot of the message showing risky elements with visual highlights.

4. AI Explanation Screen
Plain-language reasons + confidence level + model version.

5. Positive Confirmation
“Thanks for reporting” or “Marked safe” feedback with optional learning loop.

All five screens are live in a clickable Figma prototype and supported by front-end mocks in React.

Process

Flowchart Structure

User Journey Map

Personas

Results & Measurable Impact

Security Team Feedback

“The UX surfaces smarter reports and filters out the noise. We spend less time triaging and more time preventing.” (SecOps Lead)

My Design Leadership Approach

• Facilitated early design AI alignment with prototype-led discussions.

• Advocated for explainable AI, not just confident AI.

• Championed a single CTA over choice paralysis.

• Embedded micro-feedback (loading, undo, success) for trust and control.

• Pushed for Safe Preview, our standout innovation as a way to replace fear with curiosity.

Opportunities & Plans

• Deploy across desktop and Slack email channels.

• Integrate real-time model learning from user decisions.

• Add micro-education layer (30s phishing quiz)

• WCAG-compliant accessibility pass.

• Launch admin dashboard for live domain flagging heatmaps

"The UX surfaces smarter reports and filters out the noise. We spend less time triaging and more time preventing"

SecOps Lead

★★★★★