Decentralized Reputation Registry for AI Agents
Web3 ▫️ Decentralized AI Design ▫️ Front-end Development
PROJECT OVERVIEW
RelAI is a Web3-powered system that establishes a decentralized, trust-maximizing reputation registry for autonomous AI agents. It ensures that agents, whether performing code reviews, executing trades, or handling contracts, are evaluated and assigned reputations based on reliability, task completion, peer assessment, and cost effectiveness. Built during ETHGlobal New York 2025, RelAI was awarded podium placement for Best App Idea.
COMPANY
RelAI (ETHGlobal 2025 Hackathon)
ROLE
UI/UX Designer & Developer
TIMEFRAME
August 2025
TECH-STACK
Figma, GitHub (version control), Notion & Miro (planning & ideation), Spline 3D (visual design), Next.js (front-end)
THE CHALLENGE
In an agent-driven ecosystem, users struggle to trust AI agents for tasks like code review, trading, or research. Existing reputation systems are centralized, fragmented, and opaque, leaving users with confusion, decision paralysis, and security risks.
MY APPROACH
RelAI delivers a decentralized reputation dashboard that makes AI agent performance transparent. Clear metrics like task success, reliability, efficiency, and peer validation are translated into intuitive visuals and accessible flows, enabling users to quickly compare agents and choose with confidence.
STEP 1
DEFINE
UNDERSTANDING THE PROBLEM
As AI agents begin to act autonomously, handling tasks like contract negotiation, financial transactions, or technical reviews, users struggle with one fundamental question: “Which agent can I trust?”.
WHY CURRENT SYSTEM DOESN'T WORK
No insight into Agent performance, leaving users unsure of agent reliability.
Metrics from multiple sources are fragmented, making comparison difficult.
Complex AI and Web3 data lacks intuitive visualization.
Users can’t interact with or explore data to gain confidence in decisions.
STEP 2
IDEATE
SOLVING THE PROBLEM
To make AI agent reputations clear and actionable, I explored ways to visualize complex data and design intuitive interactions.
BRAINSTORMING SESSIONS
Sketched dashboards, charts, and badges to communicate performance at a glance.
Focused on interactive and engaging visualizations for trust and clarity.
Explored filters and comparison views for task-specific agent evaluation.
DEFINING REPUTATION METRICS
To calculate and display agent reputations meaningfully, I collaborated on choosing metrics that were both quantifiable and easy to understand:
Task Success Rate: How often an agent successfully completes tasks
Reliability Score: Consistency of performance over time.
Efficiency: Cost-effectiveness or resource usage per task
Peer Ratings: Community feedback from other AI agents or users
Recency / Activity: Ensuring scores reflect current performance, not outdated data
STEP 3
DESIGN
CREATING THE SOLUTION
I designed a futuristic, Web3-ready interface for a decentralized environment, using cyberpunk-inspired colors, neon gradients, and glitch effects to create an immersive, visually striking experience that communicates trust, transparency, and the cutting-edge nature of AI agents.
3D INTERACTIVE EXPERIENCE
Built a 3D blob in Spline that reacts to the user’s cursor, creating an engaging and dynamic interaction.
Designed interactions to visualize AI activity and reputation in a tangible, intuitive way.
Used cyberpunk-inspired colors, neon gradients, and glitch effects to create a modern, visually striking 3D interface that feels futuristic, AI-driven, and trustworthy in a decentralized Web3 environment.
VISUAL SCREEN DESIGNS
STEP 4
DESIGN TO DEV
APP PROTOTYPE
I translated the futuristic 3D interface and interactive dashboards into a working platform using Next.js for the frontend and Three.js to render dynamic, interactive metric visualizations. This ensured the 3D blob, cyberpunk aesthetic, and AI reputation data were fully interactive, visually engaging, and Web3-ready, bridging the gap between design concept and functional implementation.
KEY LEARNINGS
AI Design — Abstract systems like agent reputation require data storytelling, simplifying technical performance into clear, visual narratives users can trust.
Web3 Design — Designing for decentralized systems highlighted the need for self-explanatory interfaces, since there’s no central authority guiding the user.
Hackathon Pace — Rapid prototyping under tight timelines emphasized the value of prioritizing core functionality and clarity over pixel-perfection.
THE IMPACT
Earned a podium finish for Best App Prize by Zircuit at ETHGlobal New York 2025, validating both the concept and execution.
Showcased how a decentralized AI reputation registry can tackle trust and interoperability issues in Web3 ecosystems.
Reinforced my strength in end-to-end ownership, from design to development, under hackathon pressure.
CONCLUSION
Working on Curiouser.AI was a transformative experience that deepened my expertise in product design, UI/UX, and brand strategy.
