Inspiration
The inspiration for Pup AI came from the frustration many retail investors face when trying to spot promising meme coins early—especially on the Avalanche (Avax) chain. While insiders and automated bots consistently secure early entries, everyday traders are left chasing pumps and missing the real profits. We wanted to design a system that could detect early signals in the noise, give fair access to quality intelligence, and ultimately make meme-coin investing safer and more data-driven.
What it does
Pup AI is an AI-powered early detection system for Avax-based meme coins. It continuously monitors decentralized exchanges, social media chatter, and blockchain data to identify tokens with real growth potential before major price movements occur. The system goes beyond hype—it analyzes sentiment, on-chain behavior, and community authenticity to separate genuine opportunities from scams or rug pulls. Best of all, its profit-sharing model ensures that the platform only earns when its users do, creating perfect alignment of incentives.
How we built it
The entire Pup AI ecosystem is hosted on AWS EC2 to ensure scalability and uptime. The frontend, built with Next.js, manages wallet connections, user authentication, and provides a seamless interface to interact with our analytics and AI tools. A custom scraper gathers data from Dex Screener and Twitter, focusing on Avalanche-related tokens and community sentiment. This information flows into our AI analysis layer powered by AWS Bedrock, where large-language models assess risk scores, potential upside, and deliver actionable insights.
We also integrated an Eliza Agent, enhanced with Retrieval-Augmented Generation (RAG), allowing users to query the system conversationally and receive up-to-date, context-aware answers. This intelligent assistant combines blockchain data, real-time social metrics, and AI reasoning to provide clear investment rationales. Together, these components form a unified system where only verified, paying users gain access to premium, real-time intelligence.
Challenges we ran into
Building Pup AI involved navigating several technical and conceptual hurdles. Transitioning to the latest version of Eliza introduced compatibility challenges since much of its internal structure had changed, and the available documentation lacked examples for contextual knowledge integration. Additionally, learning AWS Bedrock and understanding its APIs added an extra layer of complexity. Connecting the knowledge base effectively to Eliza’s retrieval system required extensive experimentation. Through persistent trial and error, multiple test runs, and reviewing code from related projects, we eventually achieved seamless contextual knowledge retrieval and stable model performance.
Accomplishments that we're proud of
We’re proud to have built one of the first AI systems focused on early meme-coin detection specifically for the Avalanche ecosystem. Achieving accurate token risk scoring, genuine community sentiment recognition, and integrating a conversational AI layer were major milestones. Another key accomplishment was implementing a fair profit-sharing model—something rarely seen in the trading-signal space—that ensures users and the platform succeed together.
What we learned
The project deepened our understanding of large-scale data scraping, natural-language sentiment analysis, and multi-agent AI systems. We learned how to bridge on-chain analytics with social data to derive meaningful, real-time insights. Perhaps most importantly, we gained a new appreciation for the challenges of keeping AI systems up to date with rapidly changing blockchain dynamics and social trends.
What's next for Pup AI
Looking ahead, we plan to expand Pup AI’s capabilities beyond Avalanche to other emerging blockchains, add support for additional data sources like Telegram and Discord, and further refine the Eliza Agent’s reasoning abilities. Our goal is to transform Pup AI into a full-fledged, cross-chain intelligence assistant for meme-coin investors—making crypto trading more transparent, fair, and profitable for everyone.
Built With
- amazon-web-services
- css3
- javascript
- next.js
- node.js
- react
- solidity
- tailwind
Log in or sign up for Devpost to join the conversation.