Featured Case Study
A cross-platform trust assistant for suspicious AI-generated political content.
Signal helps everyday social media users identify suspicious AI-generated political content in real time, understand why it was flagged, and make more informed decisions before trusting or sharing it.
Problem
During election season, users encounter political content that may be AI-generated, misleading, or missing context. Because that content spreads quickly across social platforms, people may trust or share it before verifying it.
Signal addresses this gap by adding real-time trust guidance directly into the browsing experience, helping users pause, understand why content was flagged, and review evidence before they share.
AI-generated political posts can look credible at a glance.
Misleading content moves through communities before context catches up.
Most users do not have time to fact-check every post while scrolling.
Goal
Users should understand why political content was flagged, see supporting evidence, and stay in control of monitored platforms, alerts, and privacy.
Target User
Signal is designed for people who encounter political content while browsing platforms like Instagram, X, TikTok, and YouTube. The prototype centers on Lamar, a 26-year-old community-oriented user in Akron, Ohio who wants to stay informed during election season without becoming a professional fact-checker.
Lamar represents the core audience: thoughtful but not expert, politically aware but not professionally trained, and exposed to real-time content that shapes what people around him believe and share.
Process
Scoped the risk around AI-generated political content during election cycles.
Defined everyday social media users like Lamar as the primary audience.
Separated Signal into a standalone app and a live monitored browsing layer.
Built low- and high-fidelity Figma flows around in-feed trust labels and evidence.
Ran usability sessions to test clarity, hierarchy, onboarding, and trust cues.
Refined explanations, visual hierarchy, navigation, and exposure history.
Delivered a trust assistant concept focused on judgment support, not automation.
Research
I reframed the project around how everyday users decide whether political content is trustworthy while browsing. The product needed to support quick judgment without feeling like surveillance, censorship, or an expert-only tool.
Design Decisions
Decision
Design Signal as one product with two connected layers.
Reason
The standalone app gives users a home base, while live monitored browsing helps them in the exact moment suspicious content appears.
Tradeoff
The product model is more complex to explain than a single dashboard.
Decision
Use in-feed trust labels instead of blocking or hiding content.
Reason
Signal should support judgment, not make the final decision for the user.
Tradeoff
Labels must stay subtle enough to feel calm but visible enough to interrupt risky sharing.
Decision
Pair every flag with reasons, evidence, and a share warning.
Reason
Users need to understand whether a claim is disputed, a source is questionable, or a synthetic marker appears.
Tradeoff
Evidence screens can become dense if hierarchy and plain language are not handled carefully.
Decision
Make platform monitoring, alerts, and privacy explicitly user-controlled.
Reason
Trust guidance should not feel like secret background surveillance.
Tradeoff
More controls require careful onboarding so the experience still feels approachable.
Prototype
The core journey moves from the Signal app into live monitored browsing: the user opens a supported platform, browses normally, encounters suspicious political content, sees an in-feed trust label, reviews an explanation and evidence, and pauses before sharing.
The embedded Figma flows show how Signal moves from early structure into a high-fidelity trust assistant with live browsing, in-feed labels, evidence drawers, and user-controlled settings.
Usability Testing
I ran moderated usability sessions on the low-fidelity Figma flows. Participants understood that Signal helps identify suspicious political content and gives context before sharing, but the tests surfaced problems with clarity, hierarchy, onboarding, navigation, and dense explanation screens.
The concept was strong, but the interface needed to communicate trust faster. Users needed clearer value up front, more visual content previews, and stronger hierarchy around why a post was flagged.
The app felt text-heavy. Users wanted more icons, images, stronger visual cues, and real content thumbnails — especially for Instagram and Reels.
The flow was confusing. Users asked for fewer "Skip for now" outs, a clearer upfront benefit, a simpler step-by-step structure, and larger, more obvious checkboxes and buttons.
The back arrow felt unclear. Users wanted wording that showed when they were returning to the main app, and stronger labeling and hierarchy across navigation buttons.
Some controls read as decoration, not as interactive elements. Onboarding controls especially needed to look tappable, and primary actions needed to stand out more.
Users wanted clearer structure in the explanation and evidence areas, plus a faster way to scan why content was flagged.
Too much information and too much reading. Users wanted it more concise, more visual, and easier to scan quickly.
Three steps users could repeat back unprompted after the second round of testing.
The biggest single change: make trust guidance visible in the moment of browsing, then simplify the explanation screens so users can understand the flag before deciding whether to share.
Iterations
Each iteration responds directly to a pattern from the usability sessions above.
What it was: A setup flow that did not clearly explain what live monitoring did or how users stayed in control.
What testing said: Users understood the broad idea, but needed the value and privacy boundaries explained earlier.
What changed next: Rewrote onboarding around three plain-language steps: choose monitored platforms, browse with live labels, review evidence before sharing.
What it was: Text-heavy screens explaining trust labels and flagged content in long paragraphs.
What testing said: The app felt dense; users wanted icons, content thumbnails, and visual cues over walls of copy — especially for Instagram/Reels.
What changed next: Replaced paragraph-led screens with thumbnail-first cards, iconography for flag categories, and reduced body text by roughly half.
What it was: A bare back arrow, low-contrast buttons, and onboarding controls that didn't look interactive.
What testing said: Users weren't sure where the back arrow led, mistook buttons for labels, and missed primary actions.
What changed next: Labeled the back arrow ("Back to feed"), promoted primary actions visually, and gave every interactive control a button-shaped, high-contrast treatment.
What it was: A single trust/evidence screen with mixed information types, and an exposure history list that read like a log.
What testing said: Users wanted clearer sections for disputed claims, source issues, synthetic markers, and context.
What changed next: Introduced a more structured evidence layout and rebuilt exposure history into a compact, visual review surface.
Outcome
Signal delivers a high-fidelity concept for a cross-platform trust assistant. Instead of hiding content or making decisions for the user, Signal adds trust labels directly in-feed, explains why posts were flagged, provides evidence before sharing, and gives users control over monitored platforms, alerts, and privacy.
Reflection
Designing for trust means designing for agency. Signal cannot feel like it is policing content or secretly watching users. It has to explain what it is doing, show why something was flagged, and leave the final decision with the user.
The strongest version of the concept emerged when the product moved closer to the moment of risk: live browsing, in-feed labels, and evidence before sharing. The standalone app still matters, but the trust layer becomes most useful when it appears in context.
Next Steps
Keep exploring
Return to Monoscribe for the full hardware-software product case study, or get in touch about product design roles.
