Civica uses AI to help people understand and respond to government actions that impact them
Summary
A 10-week senior thesis research project exploring how human-centered design and AI can help marginalized communities understand and respond to government regulatory changes that may affect them.
Developed a mobile app prototype that demonstrates how AI can make government regulations more understandable for marginalized communities. The research findings will compete for publication in an academic journal at TAGA in 2026.
Read the case study ↓
Context
Why this project?
I recognized a critical challenge people are facing in navigating constantly evolving government regulations at the local, state, and federal levels. I got curious about leveraging AI to transform overwhelming regulatory complexity into clear content and actionable steps for people to follow.
I hypothesized that...
Discovery
Most want to be informed, but few trust how
I conducted a survey to gain quantitative metrics about people's news engagement habits and experiences.
The survey included 110 participants from various backgrounds and highlighted that while existing delivery methods of news articles are effective and people want to be informed, they struggle to trust news and media outlets to inform them.
of respondents’ primary barrier to engaging is a lack of trust in news
of respondents find reading articles or reports to be most effective
of respondents want to be more informed about what impacts them
Target Audience
Who most critically needs to be informed of upcoming policy changes?
Through researching the 2024 GOP platform and Project 2025, which outlined proposed policies and changes to government structures, I identified 3 user groups that need support:
⚧
Transgender individuals
Addresses urgent healthcare and legal barriers they face due to discriminatory policies
🪪
Undocumented Immigrants
Supports challenges of access to essential services from restrictive immigration laws
🫄
Reproductive care seekers
Responds to the increasing threats against reproductive autonomy and access to care
While everyone in America deserves to understand how policy changes impact them, to maintain scope, I decided to center the most impacted communities now to create a strong foundation that can be adapted to serve diverse populations later.
User research
Analysis
I conducted 6 user interviews with people from the target audience to uncover their pain points when accessing the information they need. In addition, I conducted six expert interviews with professionals across adjacent fields to ground the work in subject matter expertise. Using Dovetail, I compiled and tagged all of my interview and survey data to then create insight stories.
People problems
From those insights stories, I focused on the top 3 people problems.
🤨
Mistrust of information
Suspicion of government, technology, and news has resulted in mistrust and fear in people, something exacerbated for marginalized people with systemic trauma
🤯
Information overload
With vast amounts of information scattered across multiple sources and constant media exposure, people struggle to find clear, relevant, and actionable info
🕹
Lack of user control
Mysterious algorithms make people unsure of what/how personal information is being used. Additionally, people often disengage due to notification overwhelm
Design goals
Stemming from the people problems and rooting in civic design principles, I defined 3 key design goals to guide the prototype design.
Transparency
To create trust in the platform, clearly disclose how AI is collecting and using personal data and where information is sourced from
Clarity
Present complex legal and policy information in digestible, plain language chunks with an emphasis on relevance and actionability
User agency
Allow people to decide what information they disclose, what notifications they receive, and what information is relevant to them
IDEATION
Rapid sketching
To kick off brainstorming, I did a couple of rounds of timed sketches, exploring multiple concepts for the onboarding, home, update, and saved pages.
While the onboarding screens were straightforward, leaning on familiar flows to fit people's existing mental models, structuring the policy updates required extra workshopping. I played with chat-style and tap-through formats before sticking with a summarized article-like format that aligned with my earlier finding that 75% of people most effectively process information by reading articles or reports.
Design
MVP Prototype
Guided by "how might I?" statements, I conceptualized solutions to my design goals and created a minimum viable product prototype.
Transparency
How might I create transparency to build trust?
🔏
Disclosures
Disclosures appear during onboarding before asking for personal information
The AI Disclosure explains where information comes from, how it’s delivered, and acknowledges possible biases and inaccuracies.
The data disclosure explains how people’s personal data is being used, reassuring them that the app maintains their privacy, and directing them to where they can control what data is being used.
📜
Context for every question
Information asked during onboarding can be sensitive so each question includes a brief explanation of how it connects to relevant policies, making the purpose clear and helping people understand if the question is relevant to them
Clarity
How might I break down complex legal and policy information clearly?
📝
Plain language
As one interviewee put it, legal language often feels “as opaque and as verbose as possible,” leading me to follow plain language guidelines:
Personalized headlines that highlight why something matters to the individual
Question-based follow-ups make it easy for people to get additional relevant information with progressive disclosure, if needed
Everyday language that limits jargon
List formatting to answer complex questions in quick, digestible chunks
👉
Actionable next steps
Users reported helplessness when reading news because they didn’t know what to do next. The last portion of the policy update features contextual next steps tailored to the issue at hand, helping users move from awareness to informed engagement
User control
How might I balance user control with personalization?
⏩
Skippable questions
When the app asks for personal information, people can skip the questions with an option to revisit their choices later
🔔
Tiered notification control
Many people reported feeling overwhelmed by the number of notifications they receive. They also had differing opinions on what information they wanted to be notified about.
Tiered notification enables people to choose how often they receive updates and the content of those updates
🔁
Relevance feedback loop
To ensure that content is relevant and critical, people can select a thumbs up or down for each update, so AI can better personalize the content they receive
Impact
Usability testing
I ran virtual usability sessions with participants from the initial interviews, using the think-aloud method to understand their thought process as they navigated the app.
People liked the progressive disclosure and the actionable steps that made policy feel less overwhelming. At the same time, they flagged places that needed more clarity—like text hierarchy, question wording, and making clickable items obvious. Concerns about AI bias and how info is verified highlighted that the product needs more transparency features.
Conclusion
This project confirmed that people, especially those in marginalized communities, need government information that’s clear, trustworthy, relevant, and actionable. Due to the short timeline, the prototype is unfinished, made with the research purpose of testing the concept. The prototype can’t prove impact yet, but it validates that AI can help make government complexity understandable. In the future, I hope to refine and complete the prototype to include user feedback and create a completed prototype.
Reflection
Lessons learned
This project was the hardest yet most rewarding project I’ve worked on.
One of my largest challenges was scope creep; I wanted to take on the world, but tackling such a large project in a short amount of time made reducing scope important to deliver on my goals and keep to my schedule. Because I wasn’t working with a team, I had to learn how to make quick decisions and hard tradeoffs, often without the input of others. This taught me how to think critically and step back to consider the bigger picture.
I wore many hats but it wasn’t a one woman show, I learned how to ask for guidance when I was out of my depths. Leaning on people who were experts in their field, I was able to take the project farther down roads I didn’t even know existed. Thank you to everyone who led me down them.