Mobile
Desktop
Data-Driven
Product Thinking
Solutions
Demo Mode
Users will be able to preview the conversation with complete info (placeholder) to see what's
the conversation look like
Finanacial Profile
Users will be able to input their financial information so that the chat bot can provide a more credible and precise answer. The chatbot will also collect users' info during conversation and store them in the financial profile.
Information Cards
Users will be able to see cards that related to the topic of the conversation so that they can have a better understanding
or fill in the incomplete information to make the answer more credible and authentic.
© Testimonials レビュー
(WDX® — 06)
Real Feedback
Standing My Ground Against “Perfect Data”
The moment that stayed with me most was the heated debate with our CEO over whether users should be forced to complete every financial card before they could even start talking to the chatbot.
On paper, “perfect data” sounded like the safest way to guarantee accurate answers, but everything I’d seen from our previous product told me that gating the experience behind forms would feel more like an exam than a conversation. I carved out time for quick user tests comparing a hard‑gate flow with a more flexible, “good enough” path, and when people naturally chose the latter, it gave me the evidence I needed to push for a compromise where users could ease in, build trust, and fill in details over time.
That moment taught me my job wasn’t just to sketch nicer flows, but to create enough proof for the team to believe that a more human, conversation‑first approach could still meet our business goals.
My Role
UX Design Intern
Duration
May-August 2025
Tools
Figma,Photoshop
Capcut
Team
UX Designers,PM
Engineers
Impact
80%
Reported feeling less pressured with progressive disclosure
40%
Reduction in cognitive load after restructuring financial cards into progressive inputs.
25%
Reduction in time-to-first-response.
From Low UX Influence to Quietly Steering a Constrained AI Product
Story Part I-Refusing to Be “Just the Person Who Draws Screens”
Kickoff & Alignment · Product Sense · Ownership Mindset
When I joined the Wefire chatbot project, the brief sounded simple: rebrand an AI finance chatbot and make adjustments that fit the old architecture and tight budget. Sitting in those kickoff meetings with a CS‑background CEO and PM, I realized how easy it would be for my role to shrink into “just draw whatever is feasible,” and that was exactly what I didn’t want. So I set a different goal for myself: treat every UX decision as a product decision, and use each alignment meeting to ask “why” and articulate trade‑offs, even when I wasn’t sure my voice would carry as much weight as the engineers’.
Story Part II-Turning “No Research Budget” Into My Own Discovery
UX Audit · Competitive Analysis · Problem Framing
Very early on, I learned there was no time budgeted for user interviews, usability testing, or even access to analytics, yet we were still expected to move straight into design. Instead of waiting for research that would never come, I audited the previous Wefire flows, studied competitors like Perplexity and ChatGPT, and mapped out where the chatbot’s credibility would break once it was separated from users’ financial data. It was uncomfortable knowing that many decisions came from my own audit rather than clean research reports, but it also forced me to frame the problems myself and use that structure to guide our design direction, rather than just reacting to ad‑hoc requests.
Story Part III-Turning Pushback and Budget Constraints Into a Win
for Users
Presentation · Dev Trade‑off Negotiation · Influence Without Authority
The toughest moments came during design reviews and handoff: with a tiny budget, engineers pushed back hard on many decisions—side menus, demo mode tags, even motion—and “let’s just stick to the old Wefire design” became a common refrain. I was frustrated that my rationale for a more conversational, optional financial card flow wasn’t convincing enough on its own, so I carved out extra time to run quick user tests and used the results to negotiate a compromise that still let users ease in, get answers, and fill in details later. During delivery, I also learned to work with the CEO as an ally to unblock feasibility issues and simplify interactions without losing the core UX patterns, which was the first time I felt I could influence a strongly opinionated, budget‑driven team without having formal authority.
