Yidan Gong Portfolio

Designed the core user experience for a conversational AI platform, shaping conversation, feedback, and progress flows.

Year

2024

Company

Articulate.AI

My Role

Founding Product Designer

Conversation Designer

Target Users

Adult learners looking for a more guided and interactive way to practice real-world communication

Team

Founders

Engineers

AI Scientists & Researchers

Context

A trained AI model built for conversation needed a product experience around it

At Articulate.AI, the engineering team had already built the core conversational AI capability, my responsibility was to turn that capability into a clear product experience people could use.

Context

A trained AI model built for conversation needed a product experience around it

At Articulate.AI, the engineering team had already built the core conversational AI capability, my responsibility was to turn that capability into a clear product experience people could use.

Context

A trained AI model built for conversation needed a product experience around it

At Articulate.AI, the engineering team had already built the core conversational AI capability, my responsibility was to turn that capability into a clear product experience people could use.

Unclear product vision

Constrained engineering bandwidth

Need to validate user demand

My responsibility: build a cohesive and usable product experience

Target users

English learners with different communication goals

Whether users were building confidence for everyday communication or preparing for more structured goals like exams, the product offered a more interactive and guided experience.

Target users

English learners with different communication goals

Whether users were building confidence for everyday communication or preparing for more structured goals like exams, the product offered a more interactive and guided experience.

Target users

English learners with different communication goals

Whether users were building confidence for everyday communication or preparing for more structured goals like exams, the product offered a more interactive and guided experience.

Everyday communication

Practicing more flexibly

Improving through feedback

Preparing for exams

Research

Mapping communication practice journeys to uncover where AI could create value

I conducted 13 user interviews, field observations, and competitive analysis to understand how adults practiced communication across structured learning, self-study, and real-world use. My goal was not simply to turn pain points into a smoother flow, but to identify where AI could create a more interactive and feedback-rich product experience that users would want to return to over time.

Research

Mapping communication practice journeys to uncover where AI could create value

Research

Mapping communication practice journeys to uncover where AI could create value

Problem

English learners struggled to use what they learned in real-world communication

I defined the core product direction through user research by clarifying the user pain point and identifying the key breakdowns that would shape the design.

Problem

English learners struggled to use what they learned in real-world communication

I defined the core product direction through user research by clarifying the user pain point and identifying the key breakdowns that would shape the design.

Problem

English learners struggled to use what they learned in real-world communication

I defined the core product direction through user research by clarifying the user pain point and identifying the key breakdowns that would shape the design.

Fragmented Progress Signals

Hard to stay oriented and know what to do next

Rigid Learning Paths

Not adaptable to learner’s goals and real-life needs

Limited Real-World Practice

Practice does not reflect real communication situation

Define

Defining the core capabilities of conversational AI in the product

I worked closely with machine learning engineers to define the core capabilities conversational AI could reliably support, then mapped how those capabilities could address the key breakdowns in the experience.

Define

Defining the core capabilities of conversational AI in the product

Define

Defining the core capabilities of conversational AI in the product

Context Management

Maintains user goals, practice context, and prior conversational state.

Interaction Handling

Supports text and voice within the same conversational flow, enabling more natural turn-taking and responsive practice.

Input Interpretation

Interprets user input in context so the system can determine how to respond, guide the interaction, and support the next step.

Design

Dashboard for progress and continuity

Design

Dashboard for progress and continuity

Design

Dashboard for progress and continuity

Brings progress, active work, and next steps into one place.
Helps users understand where they are and what to do next.
Makes growth more visible over time.

Problem addressed

Progress was hard to see, and it was easy to lose track of where to continue.

Goal

Increase return usage by making progress and next steps easier to understand.

TOPIC

Travel and Directions

Current Lesson: Subway

20%

35 min left

Resume Conversation

Health and Wellness

Practice real-life conversations about staying healthy, asking for help, and feeling your best

Health and Wellness

Practice real-life conversations about staying healthy, asking for help, and feeling your best

TOPIC

Travel and Directions

Current Lesson: Subway

20%

35 min left

Resume Conversation

Health and Wellness

Practice real-life conversations about staying healthy, asking for help, and feeling your best

Health and Wellness

Practice real-life conversations about staying healthy, asking for help, and feeling your best

TOPIC

Travel and Directions

Current Lesson: Subway

20%

35 min left

Resume Conversation

Health and Wellness

Practice real-life conversations about staying healthy, asking for help, and feeling your best

Health and Wellness

Practice real-life conversations about staying healthy, asking for help, and feeling your best

Design

Goal-based topic entry

Design

Goal-based topic entry

Design

Goal-based topic entry

Organizes practice around familiar topics users can immediately relate to.
Helps users start from what feels relevant instead of entering a one-size-fits-all flow.
Makes the experience easier to approach and more flexible across different goals.

Problem addressed

Users had different communication goals, but the experience started from a fixed path.

Goal

Improve activation by making the entry point feel more relevant from the start.

Design

Scenario-based practice entry

Design

Scenario-based practice entry

Design

Scenario-based practice entry

Moves users from a topic into a specific real-life situation.
Makes practice feel more focused and easier to start.
Reduces hesitation through guided prompts before interaction begins.
Supports both text and voice input once the scenario is set.

Problem addressed

Broad topics alone were not specific enough to help users enter focused, real-life practice.

Goal

Increase practice starts by turning general topics into clearer, more actionable scenarios.

Design

A guided, customizable conversation flow

Design

A guided, customizable conversation flow

Design

A guided, customizable conversation flow

Gives users a clear agenda before practice begins.
Makes activities easier to understand through examples, tips, and action prompts.
Lets users adjust the lesson flow based on their needs in the moment.

Problem addressed

Open-ended practice made it hard to know what to expect or how to move through the session.

Goal

Increase completion by making the flow more structured, transparent, and easier to personalize.

Design

Different activity formats supported both applied communication and focused comprehension.

Design

Different activity formats supported both applied communication and focused comprehension.

Design

Different activity formats supported both applied communication and focused comprehension.

The experience included multiple activity formats to support different kinds of practice.
Some activities focused on applied communication, while others supported comprehension and reinforcement.
Together, these formats made practice feel more varied, engaging, and better matched to different learning needs.

Problem addressed

Practice could feel repetitive and did not always support different kinds of learning in the same flow.

Goal

Increase engagement and session completion through a more balanced mix of activity formats.

Speaking Role-Play

Matching Exercise

Design

Contextual Feedback

Design

Contextual Feedback

Design

Contextual Feedback

Highlights the exact part that needs attention.
Explains the correction with clear examples and simple reasoning.
Turns correction into a next step instead of stopping the interaction.

Problem addressed

Feedback was often delayed, unclear, or disconnected from the moment of practice.

Goal

Increase perceived value and engagement by making feedback more timely and actionable.

Design Process

How did I understand users and identify the opportunity for conversational AI?

I used research to separate general learning pain points from the specific places where conversational AI could create real product value.

Design Process

How did I understand users and identify the opportunity for conversational AI?

I used research to separate general learning pain points from the specific places where conversational AI could create real product value.

Design Process

How did I understand users and identify the opportunity for conversational AI?

I used research to separate general learning pain points from the specific places where conversational AI could create real product value.

Literature Review

I reviewed research on language transfer, speaking anxiety, feedback, and motivation. This showed that the challenge was not only learning English, but using it confidently in real communication.

Competitive Analysis

I analyzed English learning products and conversational AI tools to compare how they handled practice, feedback, and progress. Most felt either too rigid or too open-ended.

Semi-structured Interviews

Through learner interviews, I found recurring pain points around rigid paths, unclear progress, and practice that felt disconnected from real-life communication.

Field Observation

I observed how learners practiced across classes, self-study, and real-life situations. This helped me see where communication practice broke down in everyday use.

Design Process

How did I turn the opportunity into a buildable product direction?

This process helped me turn early ideas into a product direction that was validated through testing, aligned across stakeholders, and grounded in technical feasibility.

Design Process

How did I turn the opportunity into a buildable product direction?

This process helped me turn early ideas into a product direction that was validated through testing, aligned across stakeholders, and grounded in technical feasibility.

Design Process

How did I turn the opportunity into a buildable product direction?

This process helped me turn early ideas into a product direction that was validated through testing, aligned across stakeholders, and grounded in technical feasibility.

Rapid prototyping, testing and refining

I used quick prototypes to explore ideas early and validate what was worth developing further.

Aligning stakeholders

I brought stakeholders together around shared priorities, tradeoffs, and product direction.

Defining AI capabilities with ML engineers

I worked with ML engineers to clarify what conversational AI could reliably support and translate those capabilities into product direction.

Prioritizing features

I narrowed the experience to the most valuable features first, balancing user impact, product goals, and implementation scope.

Supporting engineering development

I collaborated closely with engineers through handoff, design specs, and interaction reviews to accelerate MVP development. We aligned on implementation details to build the experience clearly and consistently.

Supporting leadership in investor presentations

Beyond product design, I supported leadership by creating product visuals and demo materials, helping frame the product narrative around business value and communicate its potential to investors.

Iterations

Reframed a linear flow into a scenario-based experience, achieved ~20% improvement in early continuation

I led the design and iteration of conversation flows, using concept testing to shift from a fixed, linear structure to flexible practice paths aligned with learners’ goals.

Iterations

Reframed a linear flow into a scenario-based experience, achieved ~20% improvement in early continuation

I led the design and iteration of conversation flows, using concept testing to shift from a fixed, linear structure to flexible practice paths aligned with learners’ goals.

Iterations

Reframed a linear flow into a scenario-based experience, achieved ~20% improvement in early continuation

I led the design and iteration of conversation flows, using concept testing to shift from a fixed, linear structure to flexible practice paths aligned with learners’ goals.

Iteration 1 : Linear Sequence

Concept Testing Findings:

Practicing all 4 skills in one topic was overwhelming.

Topics were engaging, but learners wanted more depth and real-life scenarios.

Desire for flexibility to choose or switch skills during a session.

Iteration 2 : Topic Focused Practice Flow

Canvas

100%

Canvas

100%

Iteration 2 : Exam Focused Practice Flow

Canvas

100%

Site Map

Canvas

100%

Iterations

What user feedback revealed about the feedback experience

I collected user feedback on the early feedback experience and found that correction alone was not enough. Users needed clearer explanations, more guidance, and a better sense of what to do next.

Iterations

What user feedback revealed about the feedback experience

I collected user feedback on the early feedback experience and found that correction alone was not enough. Users needed clearer explanations, more guidance, and a better sense of what to do next.

Iterations

What user feedback revealed about the feedback experience

I collected user feedback on the early feedback experience and found that correction alone was not enough. Users needed clearer explanations, more guidance, and a better sense of what to do next.

Feedback felt too mechanical and easy to move past, making it harder for users to feel supported or act on it.

I redesigned the feedback to work at multiple levels: pinpointing the exact error inline, offering a lightweight hint, explaining the rule in context, and prompting the user to retry right away.

Different types of corrective feedback EduBot gives:

After doing additional secondary research and studying how feedback is delivered in real time, I developed a feedback framework to guide how the AI responded during practice. I also worked closely with ML engineers to define how different feedback types should be delivered in the product. The table below shows when each type of feedback appeared and how it was used.

Iterations

How usability testing shaped the design

I planned and conducted usability tests, analyzed the findings through affinity mapping, translated the insights into design updates, and shared the results with the team.

Iterations

How usability testing shaped the design

I planned and conducted usability tests, analyzed the findings through affinity mapping, translated the insights into design updates, and shared the results with the team.

Iterations

How usability testing shaped the design

I planned and conducted usability tests, analyzed the findings through affinity mapping, translated the insights into design updates, and shared the results with the team.

Moved the primary action into the left navigation so users could start a new conversation more easily.

Added icons before lesson content, hints, and bot prompts to separate each type of information visually.

Added dynamic underlining so users could follow the text as it was spoken.

Reflection

What this project strengthened in me

More than a product outcome, this project pushed me to grow in judgment, collaboration, and the way I design under complexity.

Reflection

What this project strengthened in me

More than a product outcome, this project pushed me to grow in judgment, collaboration, and the way I design under complexity.

Reflection

What this project strengthened in me

More than a product outcome, this project pushed me to grow in judgment, collaboration, and the way I design under complexity.

I grew into end-to-end product thinking

Being the only designer meant I had to think through the whole product, not just individual screens. I learned how to connect research, priorities, and design decisions into one direction that could actually move the product forward.

Stronger prioritization under constraints

This project strengthened my ability to focus on what mattered most when time, resources, and technical feasibility were all limited. I became more intentional about making tradeoffs instead of trying to solve everything at once.

I became more effective at cross-functional alignment

Without another designer to lean on, I had to get much better at explaining my thinking and bringing people along. Prototypes, conversation flows, and testing insights became the tools I used to create shared understanding across the team.

More confidence in shaping ambiguity

Working in an early-stage environment pushed me to make progress without waiting for perfect clarity. I learned how to turn open questions into a clearer direction through research, synthesis, and fast iteration.

A deeper sense of what makes an experience stick

I became more aware that a product is not only about core functionality, but also about whether people feel guided, supported, and motivated to come back. That shifted how I think about continuity, tone, and overall product value.

A sharper sense of AI product judgment

This project pushed me to think more carefully about what AI should do, and what it should not. I became more aware that designing for AI is not just about capability, but also about trust, tone, and how supported the user feels.