Clinc
2024.9 - Present
B2B SaaS | Product | UX UI | Agile
My Role
Product Manager (From 2025.4)
Clinc is a vertical SaaS platform that enables conversational designers inside banks to build their own customized chatbot experiences—powering U.S. Bank, No. 1 digital banking assistant with 51 million+ queries each year.
As the UX and product designer, I led 15 initiatives (6 new features and 9 enhancements, 6 launched live) to streamline workflows and improve usability.
Revamping Clinc’s Conversational AI Platform through a User-Centered Approach
Case Study
Slot Mapping Page Redesign
My Role
User Research, User testing, UX/UI Design, Product management
Team
CTO​, Product Manager, Conversational Designer, Project Manager​
Time Frame
6 weeks
Impact
>40% faster editing experience (estimated by number of clicks)
100% Positive internal feedback
Framework adopted for future design guidance
Why We want to Redesign?
The Slot Mapping page helps designers build a clear taxonomy by linking synonymous user inputs to one standardized value, ensuring consistent interpretation by the NLP model.
For Instance
Huge
Big
Gigantic
will be mapped to
Large
However, the existing workflow was inefficient, including:
​
-
Too many steps to edit
-
Not enough working space
-
Repeated and manual work
-
No flexibility for iteration
As a result, the process became time-consuming and disrupted designers’ focus during model training.
Goals
How might we help conversational designers iterate faster and work more efficiently while preserving familiar workflows?
Understand User Behaviors through Research
To uncover workflow pain points, I conducted two rounds of qualitative research with internal conversational designers. The goal was to understand how mapping is performed today, where time is lost, and what slows down iteration.
Broad Exploration
Figure out how designers currently create, edit, and test slot mappings.
Targeted Validation
Validate pain points and test early hypotheses of specific features.
Key Insights from Interviews
I synthesized interview notes and identified recurring patterns that shaped our design priorities. Designers consistently reported pain points around workspace efficiency, discoverability, and repetitive editing. The examples below illustrate these issues in the existing platform.
Limited Workspace Area
The main editing workspace uses less than 1/3 of the screen
"Whenever I need to edit the value and secondary cluster, I feel there is not enough space."
The old UX required unnecessary effort to learn
The edit icon appears visually disconnected from the table.
"It wasn’t obvious how to edit the table—the edit icon felt too far away and easy to miss."
Cumbersome editing experience
Users need to click at least 3 buttons to edit the values.
"Having to click both “Save” and “Save Mappers” to complete edits feels redundant and unnecessary."
What I Discovered from User Interviews
I synthesized notes through affinity mapping and task-flow analysis, identifying patterns that directly informed design priorities. User interviews surfaced consistent pain points around workspace efficiency, discoverability, and repetitive editing.
The analysis has been grouped by features, including pain points, goals, and possible solutions.
(Click the image to zoom in and see the details.)
How the Key Findings Informed the Design Decisions
Finding
Informed Iterations
01
Limited working area slowed review
Introduce adjustable panels and prioritized workspace around active tasks to reduce scrolling and keep users in context.
02
No clear state after actions
Added real-time action feedback so users always know what just happened.
03
Too many steps to complete changes
Simplified interaction flow by removing redundant confirmation layers and consolidating “save” actions.
04
Users had to manually recreate similar mapper structures
Added clone and import capabilities to support reuse, reducing repetitive setup effort during iteration.
Participatory Co-Design with Experts to Identify the Ideal Layout
Since many pain points directly pointed toward actionable fixes, I focused on layout and interaction explorations for workspace optimization.
​Be Able to enlarge the edit space
Two-column layout & adjustable sidebar
Collapsible component & vertical workspace
After collaborating with senior platform users, we combined the two-column layout from Wireframe 2 with the collapsible UI components from Wireframe 3, forming the foundation for the final redesign.
Final Design
Better Use of Space Aligned with User Workflow Priorities
Before
The old design used a rigid layout with limited space for intensive work.
After
The new design uses adaptive layout, expanding workspace from 30%→50%.
It minimizes low-priority panels and letting users focus on key tasks.
Clearly indicate the system Status
Before
No visual feedback after selection, causing confusion.
After
Add real-time visual feedback and highlight states to make every selection visible and confirm system response instantly
A More Seamless and Efficient Experience
How the assigning experience was improved?
Simplified the assigning flow with a more visible “Add” signifier.
"Save" that previously required two steps now complete in one, improving speed and clarity.
Resulted in reduced cognitive effort and aligned the behavior with users’ natural expectations.
Before
After
Another major pain point—the redundant editing workflow—has also been significantly improved.
Before
After
Removed unnecessary confirmation clicks — the new design supports immediate inline edits, speeding up iteration.
Reduced steps from 3 → 1, significantly shortening task time.
Freed users to focus on refining mappings instead of being troubled with UI friction.
Enhance flexibility and reduce manual work through new features
A new drag-and-drop editing system replaces static pipelines — enabling users to reorder, clone, or modify mapper types instantly.
It removes the need for manual recreation and improves editing scalability for large datasets.
As a designer, it’s always rewarding to see design work go into production and truly improve users’ workflows. Below is a launched feature that I led from discovery to delivery.
UX Enhancements Shipped
Post-Launch
Adopted by conversational designers across Clinc and U.S. Bank production environments.
Generation speed + output capacity tripled (10 → 30 utterances) per batch
"I could now generate and optimize model data effortlessly, saving hours of work."
What Real Users Say?
"Lin took the time to understand my workflow and the pain points I'd silently accumulated over time. She approached every issue with genuine curiosity and delivered thoughtful, innovative solutions I didn't think were possible."
Highlight for a Launched Feature
AI Generation – UX Improvements
My Role
Led end-to-end UX—from identifying pain points to redesigning the interaction model and validating with real users. Collaborated with engineering to ship the feature to production.
Feature Overview
Conversational designers often need to generate hundreds of user utterances to train NLP models. Traditionally, this was done manually or through crowdsourcing—an extremely time-consuming process. With the adoption of AI generation, designers can now produce utterances faster, but the original experience still required excessive manual editing.
Goal
Redesigned the AI utterance generation flow to reduce manual work and make the process faster and smoother for real production use.
Key Issues in the Previous UX
1
2
3
After "Generate"
01
The number of generated utterances was limited to 10.
02
Unclear add/remove interaction.
03
After generating utterances, the dialog remained in a vertical layout, forcing users to scroll and navigate back and forth.
Impact
Launched and adopted by internal designers across production environments
Generation speed + output capacity tripled (10 → 30 utterances) per batch
Feedback from End Users
"Generate Utterances tool, something that was once so cumbersome and rarely used, became intuitive, time-saving, and highly effective. I could now generate and optimize model data effortlessly, saving hours of work."
Reflection
During my time at Clinc, I grew not only as a designer but also as a product thinker and collaborator.
​
-
UX Leadership – Led end-to-end UX initiatives from discovery through delivery. I collaborated closely with the CTO, engineers, and real users, building a stronger understanding of domain complexity and technical constraints while shaping product decisions rather than just designing screens.
​
-
Product Ownership Mindset – Took initiative in defining product strategy, writing acceptance criteria, clarifying edge cases, and drafting feature documentation. I frequently led product discussions to align stakeholders, uncover risks, and drive clarity in execution.
​
-
Opportunity for Growth – I want to deepen my involvement in post-launch evaluation by gathering behavioral insights and long-term user feedback to drive continuous improvement after release.