Explore how AI accelerated UX research and design to address a fragmented Zelle fraud reporting system by, creating a transparent, user-centered resolution flow.

TLDR: AI as My Design Partner

This case study examines how AI supercharged the full UX process to address the fragmented experience Zelle fraud victims face across major banks.

By pairing AI-driven analysis with human-centered design, I uncovered pain points in fraud reporting and shaped opportunities for a transparent system. Rather than treating AI as a novelty, I showcase how it accelerated research, broadened ideation, and shortened the path from insight to prototype.

What you’ll see here:

• Synthesizing real interviews into key themes, golden nugget insights, and opportunities

• Generated personas and journey mapping grounded in real data

• Turning insights into design opportunities

• User flows, wireframes, and prototypes with AI support

• A transparent, user-friendly fraud reporting solution

The result: Faster iteration, sharper opportunities grounded in real data, and a concept that makes fraud resolution more transparent and trustworthy.

1. The Context

The challenge: Victims of Zelle fraud face a fragmented, stressful reporting flow across banking systems.

• 420k complaints at Chase, totaling $360M

• 210k customers at Bank of America, losing $290M

• 280k people at Wells Fargo, losing $220M

Cites: CBS News, American Banker, Thomson Reuters Practical Law

Why I chose this:

• A familiar, high-stakes fintech UX problem

• A chance to show how AI can accelerate research → design → prototype

2. My Human + AI Approach

AI supported me end-to-end, helping speed up research, insights, and design iteration:

• Research simulation → Personas, journey maps, and clustered pain points

• Insight translation → Turning findings into design opportunities

• Design iteration → User flows, wireframes, and click-through prototypes

3. Deep Dive: Research with AI

PSA: AI never replaces real users. But it can accelerate synthesis when exploring known problems.

When to use AI-augmented research:

• Fast patterns or direction-setting

• Early-stage hypothesis testing

• Synthesizing existing transcripts or notes

When to use real research:

• Emotional nuance & empathy

• High-risk or sensitive features

• Validation with stakeholder confidence

For this project, I leveraged real interview transcripts but used AI to speed up clustering and synthesis. Instead of weeks of recruiting, interviewing and synthesizing I spent less than a day!

• I used kome.ai to pull transcripts from 7 YouTube videos interviews of real victims of Zelle fraud.

• Prompted ChatGPT with a structured format (Role, Task, Context, Constraints, Format)

• Extracted highlights → clustered into key themes → each tied to a golden nugget insight + actionable opportunity

Grounded in real data and analysis, AI assisted me in the creation of :

• User Personas (aggregated from themes, quotes and real data)

• Journey Map grounded in actual screenshots of the Chase and Zelle fraud reporting flows

4. From Insight → Opportunity

Using the key themes, I asked AI to reframe pain points into “How might we…” design opportunities.

It generated 10, which I then grouped into 4 higher-level opportunity areas for easier stakeholder prioritization.

My AI teammate then assisted in the creation of a prioritization matrix (Impact vs Effort) to help me see which concepts might be “quick wins” vs. “longer bets.”

I was impressed with how AI cut days of synthesis into hours... turning raw interviews into clear insights ready for design opportunities. I was already seeing transparency for users just on the horizon!

5. Design Exploration with AI

Now we have arrived to the fun part!

✨ Our chosen opportunity area: Transparency & Trust ✨

Concept: “Visible Audit Trail”

• Timeline view of all user + system actions (e.g., report filed, case opened, investigator note uploaded, status updated).

• Reduces “black box” perception, increases trust.

Success metrics:

• % of users revisiting audit trail

• Fewer escalations requiring supervisor callbacks

AI generated the narrative of a step-by-step user flow for “checking fraud status” across multiple entry points. I refined the flow in Figjam for clarity and stakeholder presentation.

6. Prototyping with AI

I translated the flow into wireframes and a clickable prototype.

• Started with J.P. Morgan Salt Design System for consistency

• Used Mobbin for UI references

• Leveraged Figma Make for first-draft wireframes

• It took 81 versions… but there we are! 😅

Prototype path: 

Transaction Summary → Fraud Case Details Modal → Full Fraud Case Details Page

Ultimately, this is where I find that AI is great as that coworker who gives amazing suggestions backed with their own experiences and reasoning - but as a UX designer who is familiar with stakeholder decisions, existing design patterns, and historical reasoning, I can step in and do what human designers do best!

Try it out! Clickable desktop web prototype

www.figma.com

Created with Figma

Note: Not every button or link is interactive. Stick to viewing the status of a fraud case to see it in action! Does not work on mobile screens.

7. The Future Impacts from Testing

With AI as my teammate, I am looking forward to faster and sharper outcomes with testing.

Predicted outcomes:

• ~30% fewer customer service calls

• Reduced user stress via granular fraud statuses

• Greater trust through visible transparency

8. Results & Lessons Learned

This case study demonstrates real practice with AI in UX:

• AI as a partner for research & synthesis

• AI to supercharge design exploration

• AI to accelerate prototyping & testing

AI can cut weeks of effort into days… but it’s the human designer who ensures clarity, empathy, and trust.

Everything is moving SO fast… capabilities changed even in the 4 weeks of taking this bootcamp. I am looking to how we can further embed AI into our processes so that we can free up some time to do what humans do best… enjoy this gift of life! ❤️

Thank you for reading along and big thanks to Zander Whitehurst & Carolina Vidal of Memorisely for their knowledge sharing and excitement.

Disclaimer: The thoughts shared in this post are solely my own and do not represent the perspectives of my professional relationships, Chase, Zelle or clientele.