Enhancing AI tools to improve complex workflows and increase adoption in finance
This project reimagined the design of a comprehensive GenAI tool to help Moody’s employees work more efficiently. The legacy tools suffered from poor usability, inconsistent interactions, and limited personalization, leading to a sharp decline in adoption after launch. As the lead UX designer, I aimed to redesign this LLM-powered AI tool to enhance usability, drive adoption, and introduce a more intuitive, personalized experience for all 14,000 employees worldwide at Moody's. At the same time, the redesign set a new benchmark for AI design, ensuring a cohesive and scalable brand experience across the organization’s products.
My Role
Associate Director of Design, co-leading design within a cross-functional team of 20+ product and engineering members.
Product strategist, developing a design roadmap and defining key priorities for the MVP.
Skills
Concept Generation, User Flows
User Research, Usability Testing
Wireframe Prototyping
Sprint Management, Iterative Design Process
User Acceptance Testing (UAT)
Timeline
November 2024 - March 2025
Tools
JIRA, Confluence
Figma, Miro
Great Questions (user research)
Legacy Chat Experience
Inconsistent design, confusing interactions, usability issues
Enhanced Chat Thread Experience
Ability to navigate chat history, consistent mental models and design system
Chat Thread Feature Callouts
Navigate recent chat threads
Browse and filter all past chat threads
Legacy Files Management Experience
Overload of functionality, complicated navigation
Enhanced Files Management Experience
Consistent notification design, guided interactions
File Management Feature Callouts
Upload and organize hundreds of files at once
Secure file sharing for teams and individuals
File and folder management with real-time status tracking
Error notification and resolution system
Key Enhancements
Legacy Siloed Homepage
Complicated navigation, information overload, lack of hierarchy
Enhanced Unified Homepage
New features to enhance ease of use, improved navigation, clarity in design
Homepage Feature Callouts
Clear central entry point for user input
Multi-LLM chat platform with AI skill switching
INTRODUCTION
Project Context
To bring this vision to life, we needed to rethink how AI could seamlessly integrate into employees' workflows. The existing tools lacked usability, cohesion, and personalization, leading to low adoption rates. Our challenge was not just to redesign an AI tool but to create an experience that was intuitive, efficient, and scalable. By focusing on user needs and design consistency, we aimed to transform AI into a powerful, everyday assistant while setting a new standard for AI-driven experiences within the organization.
Our Objectives
✅ Unify AI Tools – Merge legacy tools into a single, cohesive experience.
✅ Boost Usability & Adoption – Improve interactions, personalization, and engagement.
✅ Ensure Design Consistency – Standardize branding and expand the design system.
✅ Enhance Scalability – Create a flexible framework for future AI tools.
✅ Optimize Workflow Integration – Seamlessly embed AI into daily tasks.
✅ Set an AI Design Benchmark – Define best practices for AI-driven products.
DISCOVER
Past Research Synthesis & Competitive Analysis
As the design lead on a team of 16 product and engineering members, I guided the end-to-end redesign of Moody’s Copilot tool. I began by analyzing the competitive landscape, deepening my understanding of the product’s current state, and synthesizing past research to shape the design strategy.
Competitive analysis
Synthesis of past user research
My Takeaways
✅ Turning AI potential into a clear product strategy
When I joined this project, the chatbot had powerful capabilities but no clear roadmap or cohesive experience. I quickly realized that designing AI tools isn’t just about improving interactions—it’s about getting teams on the same page about what AI should and shouldn’t do. By driving alignment between design, product, and engineering, I helped shape a more intentional AI strategy that made the chatbot more useful and approachable.
✅ Making AI feel more trustworthy and intuitive for users
AI can be unpredictable, which makes people hesitant to rely on it. I focused on designing clear, guided experiences that helped users understand what the chatbot could do and when they needed to take control. This experience reinforced how important it is to build trust in AI—not just through transparency but by making interactions feel natural and reliable.
✅ Navigating ambiguity and aligning stakeholders
AI moves fast, and priorities changed constantly. I had to learn how to cut through the noise, synthesize feedback from different teams, and advocate for the user experience while keeping the bigger picture in mind. Finding the balance between pushing for a better design and working within evolving constraints was a huge learning moment for me—and ultimately helped make the product stronger.
I synthesized the input from our personas workshop and crafted our personas. Our usability tests later on would validate these proto-personas, establishing a reference for future products.
Expanding the Design System
As the product evolved, so did our design system. I built a more expansive and powerful component library, ensuring scalability and consistency across the AI tool. Additionally, I refined the foundational styles and color system to align with the product’s development, creating a more cohesive and adaptable design framework.
Building out the components library using the Moody's branding foundational design system
DEFINE
Future State Experience Brainstorm
Personas Workshop & Proto-Personas
First, I needed to identify the key user archetypes and their unique characteristics. I designed and led the hybrid personas workshop with the design team to gather informed insights about our primary users, highlighting critical traits to validate through user research later on. This provided a clear reference point, ensuring the team designed with these target personas in mind.
Personas workshop: archetype brainstorm and target archetype deepdives
Journey Mapping and User Flows
With a clear understanding of our goals, informed by research and educated assumptions with personas,
I mapped current-state flows to identify interactions and friction points. I then developed future-state frameworks to guide the team in brainstorming and designing an optimal user experience.
Journey map
User flow - target user
Current state flows
DEVELOP
Design Development and Wireframing
Using Moody's foundational design system and branding guidelines, I began with mid-fidelity wireframes to explore key feature interactions.
I used the Moody's styles and colors branding guidelines to build the product
Mid-fidelity wireframe development
As the designs took shape, I collaborated closely with the development team to assess feasibility and incorporate their feedback. Once the major features were refined, I integrated them into an end-to-end flow to ensure a seamless and cohesive user experience.
Agile Workflow and JIRA
I tracked my progress in JIRA and helped improve the team's agile collaboration, as they were new to design and development working together. To enhance our workflow, I proposed adding more design-development grooming sessions before finalizing designs, allowing for earlier feasibility checks before high-fidelity concepts were created.
Recognizing the differences in workflows between engineers and designers, I also optimized the JIRA ticket structure by introducing design exploration tickets in sprints before development handoff, ensuring a smoother transition from design to implementation.
Example of my design tickets in JIRA
REFINE
Final Designs and Usability Testing
Dev Hand-offs
When handing off final files to the development team, I provided detailed annotations on interactions, UI elements, and use cases alongside a walkthrough/demo, ensuring clarity on design specifications.
Additionally, I organized the Figma file by user tasks, making it more structured and easily navigable for the team.
Building out the components library using the Moody's branding foundational design system
Usabiltiy Testing
We prioritized usability testing to validate the impact of our redesign and identify gaps before launch. Given tight timelines, we opted for unmoderated testing to gather quick yet valuable insights.
Another senior designer and I led testing with 15 users across key persona groups - new, infrequent, frequent, and daily users. Using Great Questions, an end-to-end UX research platform, I managed recruitment through surveys and in-person outreach.
Usability Testing Goals:
✅ Gather feedback on key interactions and first impressions
✅ Identify gaps to address before launch
✅ Validate design impact and proto-personas
The test included a background survey to validate our proto-personas, a clickable prototype to assess usability, and a quantitative evaluation of ease of use, difficulty, usefulness, and clarity.
I used Great Questions to manage this round of usability testing
Key Insights & Next Steps
✅ Clearer Entry Point for Users
Users found the new design much clearer with the prompt input search bar centered on the page, guiding their focus. This also aligned with familiar mental models from other chat interfaces, making interactions more intuitive.
Next Steps: Keep the homepage design as is.
✅ Confusing Right Panel Reopening
Users struggled to reopen the right-side panel after closing it, as the trigger wasn’t in the same area as the close icon.
Next Steps: Improved clarity by renaming "Explore AI Skills" to "Select AI Skills", making the action more intuitive. Added hotkey shortcuts so users could reopen the panel without relying on the CTA. These quick, high-impact changes were implemented before launch.
✅ Difficulty Choosing the Right Skills
With multiple skills available, users found it difficult to determine which best fit their needs.
Next Steps: Post-MVP, refine and consolidate skills to reduce redundancy and improve clarity,
ensuring users can easily select the right tool.
Quick improvements to make to MVP before launch
TEST & ITERATE
Design QA and Product Release
Our designs took shape on the Beta platform before we release this product to the 14,000 employees at Moody's. I led the QA process, reviewing each feature, logging bugs and design discrepancies in a master excel sheet and in JIRA, and ensuring the development team could address issues efficiently.
Beta platform QA → tracking in excel → prioritizing bugs on JIRA