Project Dialogue

AI-Enabled Civic Tech Platform

0→1 Product Design Civic Tech Responsible AI

Problem

Social platforms optimize for engagement over understanding, so users needed to trust a fundamentally different interaction model built around constructive disagreement.

Team

Founder
1 UI/UX Designer
3 Engineers

Role

UI/UX Designer,
Advisory Board Member

Duration

Feb 2026 – Current

Status

Pre-Launch

Key Outcome

I replaced an explanation-heavy onboarding with a single AI practice conversation that teaches constructive dialogue through use rather than instruction.

Result: ~50% fewer onboarding steps, with users reaching the platform's core value faster.

What I Drove

Product Strategy MVP Journey Mapping Onboarding Redesign AI Interaction Design Conversation Flows Brand Identity Design System Founder + Engineer Collaboration

Product

Project Dialogue is an emerging civic-tech platform rethinking how digital systems shape connection in an age of polarization. As Product Designer and Product Strategy Advisor, I translated a broad social-impact mission into MVP user flows, practice-first onboarding, responsible AI coaching concepts, and a visual identity for a new kind of digital public space.

Tools & Skills

Figma Claude Design Competitive Analysis User Flows Wireframes High-Fidelity Prototypes

You are now viewing:

01 Solution

Our first iteration.

First iteration of the AI conversation screen used during onboarding to learn more about the user

First iteration of the home screen showing live conversations and the option to start your own

A better, healthier feed.

Meet people your algorithm hid from you through matches based on difference, not similarity
Pause, timeout, and feed limits built in for healthy habit building
AI-guided practice conversations with feedback on real communication skills, not just app engagement

Visual Identity

Bold and unforgettable, but also timeless and chic. Every typographic and color decision was made to preserve that balance.

The Dialogue
Begins Here.

Crimson Text Headlines & Display

Editorial, timeless authority — headlines that feel like a broadsheet, not an app.

Start a conversation.
Share a perspective.
Change your mind.

Jost Body & UI Text

Clean and modern, balancing Crimson's formality without feeling cold.

#FE1443 Primary

#001A78 Text Navy

#011A47 Deep Navy

#FDC301 Accent

#FFFAE5 Background

02 Research

Let's Talk Competition.

I analyzed social media and match-making apps to find gaps in how platforms support cross-perspective connection.

Common Strengths

Strong content discovery
Personalized recommendations
Low-friction matching
Interest-based communities
High engagement

Common Weaknesses

Echo chambers
Outrage-driven engagement
No exposure to differing perspectives
No tools for productive dialogue

Matching Process

Optimizes for shared interests and retention
No support for navigating disagreement

How Are People Engaging Right Now?

53% Pew Research Center, 2024

of Facebook content comes from like-minded sources

41% Pew Research Center, 2024

of U.S. adults have personally experienced online harassment

53% Pew Research Center, 2016

say political discussions online are less respectful than in-person

Key Research Finding

The competitive analysis showed that the issue was not just content moderation. The deeper problem was interaction design: existing platforms reward attention, reaction, and similarity more than thoughtful exchange. That gap is what Dialogue was built to fill.

03 Ideation

Defining the Experience

The original onboarding was built around a real need: before entering the platform, users had to understand a completely new model for constructive disagreement. But as I mapped it out, I felt the process was too long and added too much friction before users could experience any real value.

Who Are We Designing For?

Young adults burned out by hostile or shallow social interactions
People who want to discuss hard topics but avoid toxic online spaces
Users who want to build communication skills across different perspectives

Original Flow Overview

Questionnaire to personalize experience

Educational modules

3 guided conversations with other users

Enter main platform feed

How might we introduce users to a completely new conversation platform and its unique interaction model without overwhelming them with lengthy explanations?

I shortened onboarding by replacing educational modules and matched conversations with a single, low-commitment AI mock conversation, giving users an early, practical feel for constructive dialogue and their own communication style.

My Solution

Practice-First Onboarding

Short setup → 2 minute AI practice → Feedback → Platform access

Reduced planned onboarding steps by about 50%

When Business Goals Meet User Limits

The founder wanted rich, real-time engagement feedback with multiple suggestions. I preserved that goal but simplified the execution, reframing the metrics around growth instead of judgment and cutting three competing suggestions down to one.

After the onboarding process was set, I shifted my focus to what the AI mock conversation screen would actually look like. The founder had a clear vision for the features she wanted to include, making sure users were aware of how they were engaging. I wanted to preserve her goal, but also make sure we were designing for cognitive load.

I brought in Claude Design to quickly turn the founder's direction into a visual mockup. It let me iterate faster and show her what she was describing so we could visualize it together before collecting feedback and revising it.

Founder's Direction

Guided Suggestions AI Practice Partner Difference Barometer Disagreeableness Barometer Dialogue Pulse

My Concern

The barometers risked shifting focus from understanding to changing minds, and three competing suggestions pulled focus from the conversation itself.

Expert Validation

Psychology and CS experts agreed: reframe the meters around growth, not judgment, and cut to one suggestion.

My Redesign

Final redesigned practice screen with Curiosity and Collaborativeness meters in the sidebar, Shared Understanding and Emotional Tension below the chat, and a single embedded suggestion

Shifted metrics to emphasize Curiosity, Collaborativeness, and Shared Understanding, and simplified guidance into a single in-line prompt.

Conversation Awareness in the Main Platform

Once users complete onboarding, they enter the main platform. These three features work together to reinforce our core goals of safe, healthy, and meaningful conversations, helping users engage thoughtfully and stay in control.

Curiosity & Collaborativeness Meters

Update in real time after each new conversation to reflect how open and collaborative your engagement has been.

User Wellness Controls

How long have users been on the platform and are they healthily engaging?

Match of the Day

Connecting users with people that other algorithms would've never paired with them, based on difference rather than similarity.

Annotated wireframe of the main platform home screen showing Performance Metrics, live Curiosity and Collaborativeness meters, Wellness Controls, and Match of the Day — Main platform home screen showing Performance Metrics, real-time Curiosity and Collaborativeness meters, Wellness Controls, and Match of the Day. These features reinforce healthy, intentional engagement once users are past onboarding.

How Every Decision Maps Back

Here's the full picture: each stakeholder's need, and what I designed in response.

Stakeholder Need My Design Response

Users A space to discuss hard topics without feeling judged, attacked, or overwhelmed Match based on difference (not similarity), AI practice before real conversations, optional prompts, and user wellness controls

Startup A focused MVP that could explain a new interaction model quickly and reduce early drop-off Short AI-guided practice flow replacing a long education-first onboarding, cutting planned steps by ~50%

Founder A direction that preserved the long-term mission while staying buildable by a small team Core flows, visual identity, onboarding logic, timed conversations, and engineer-ready concepts

04 Impact

Changing the Conversation

Project Dialogue is pre-launch, so impact is still projected, but the research makes the opportunity clear. We're not just building another social app. We're designing against the patterns that made social media toxic in the first place.

◎

Break Echo Chambers

Cross-perspective matching, not feed-driven similarity

◎

Counter Harassment Culture

Structured flows and AI moderation for civil disagreement

◎

Normalize Respectful Discourse

Dialogue skills that carry over on and off the platform

How We'll Measure Success

Beta metrics, anchored to our three product goals:

Goal 1: Reduce Onboarding Friction

Onboarding completion rate
Drop-off rate
Average onboarding completion time

Goal 2: Improve Conversation Quality

AI suggestion acceptance rate
Conversation completion rate
Change in collaboration/open-mindedness meters over time

Goal 3: Encourage Meaningful Engagement

Weekly conversations per user
Match acceptance rate
Retention (Day 7 / Day 30)

05 Next Steps

What's Upcoming?

Completed

MVP user journey and onboarding strategy
Practice-first onboarding model
High-fidelity AI practice mockup
Brand identity and design system
Success metrics and beta validation plan

In Progress

Finalize MVP Design Handoff

Refine high-fidelity user flows, complete mobile-responsive designs, address edge cases, and prepare engineering specifications.

More Expert Input

Review assumptions with experts in psychology, political science, and computer science.

Pilot + Usability Testing

Test whether users understand the interaction model, complete the practice conversation, trust the guidance, and feel prepared to enter the platform.

06 Reflection

My Takeaways

The hardest part of this project wasn't designing individual screens, it was shaping a new conversation system, balancing AI guidance, user autonomy, transparency, and trust. I also learned to balance my own design decisions around user needs with the founder's vision, since she wanted to stay true to her original idea and I wanted to honor that while still finding ways to protect retention. It taught me how to make product decisions with incomplete information, and how much faster iteration gets when AI tools like Claude Design enter the workflow early.