ROLE
TIMELINE
TEAM
1 Chief Engineer
4 Engineers
TOOL
Figma
COMPANY
A startup specializing in AI applications
MY CONTRIBUTION
As the sole designer, I shipped ChatDoc Master, a no-code B2B AI chatbot builder.
LAUNCHED PRODUCT
Available to registered business users only.
CONTEXT
Why not use ChatGPT?
SECURITY CONCERNS
ChatGPT sends user input to third-party servers.
LACK OF SOURCE CONTROL
No way to define where the AI pulls its answers from.
NO WAY TO UPDATE SOURCES
ChatGPT isn’t made for long-term content maintenance.
OPPORTUNITY
RESEARCH
I interviewed seven business users across customer support, human resources, media, and legal teams to understand their needs when using AI for information search.
MULTIPLE SOURCES
Business users want to consolidate various sources.
MATCH TONE
Teams need responses that reflect their tone.
VERIFY ACCURACY
Accuracy is critical in high-stakes fields like law or news.
UPDATED SOURCES
For fast-paced domains like media, sources are updated frequently.
ITERATIONS
I iterated on my designs based on feedback from users and engineers, focusing on usability, flexibility, and feasibility.
ITERATION 1
VERTICAL LAYOUT
I designed a top-down flow where users could upload sources at the top and view created chatbots at the bottom.
LIMITATIONS
Became inefficient as teams added more chatbots and users had to scroll extensively
ITERATION 2
HORIZONTAL LAYOUT
I moved created chatbots into a sidebar, helping users easily switch between bots while keeping the main panel focused on uploads.
WHY THIS IS BETTER
1. Better scalability for teams
2. Reduced friction in navigation between bots
3. Clearer separation of upload vs. bot management workflows
ITERATION 1
BASIC TONE AND CONTEXT CONTROLS
I initially focused on enabling non-technical users to define their chatbot’s tone and provide context to guide its responses.
LIMITATIONS
Lacked flexibility for technical users as there were no technical settings
ITERATION 2
ADVANCED TECHNICAL SETTINGS
Working with engineers, I added advanced controls such as filtering out webpages and customizing how sources are parsed.
WHY THIS IS BETTER
1. Extended functionality for power users
2. Increased platform flexibility across different roles
I designed clickable citations that would take users directly to the exact sentence in the source document.
CONSTRAINTS
Engineers flagged this as technically complex given the backend effort required for precise sentence mapping.
I introduced a scoped-down solution: clickable page number tags that link to the relevant page in the source document.
WHY THIS IS BETTER
Balanced user trust with technical feasibility
SOLUTION
I designed a tabbed interface that allows users to upload multiple data sources within a single workflow, addressing business users’ need to work with varied and multiple files.
I introduced customization options to give users the autonomy to tailor chatbots to their specific business needs.
Clickable citations let users trace each answer back to its original source with one click, reinforcing trust and aligning with user expectations for accountable AI.
Keep chatbot knowledge up-to-date
Users can update chatbots by adding or removing sources, ensuring content stays accurate over time without needing to rebuild the bot from scratch.
IMPACT
To evaluate whether the redesign resolved pain points from the old design and validate new features, I conducted usability testing with 14 users—half testing the old design and half testing the new design.
TAKEAWAYS
Working full-time in a fast-paced startup environment challenged me to grow not only as a designer, but also as a communicator and problem-solver.
Designing AI tools required making complex technology accessible. Taking the time to understand the technical concepts myself enabled me to present information in user-friendly language for users with varying levels of expertise.
Throughout the design lifecycle, I maintained open communication with the development team to validate technical feasibility. This proactive collaboration ensured design-to-development alignment, allowing me to address potential issues early through quality assurance.
In a startup environment, constraints became opportunities for creativity. This experience taught me to find creative solutions while balancing user needs, technical feasibility, and business realities.
A big thank you to my team :)
