Finrizz

This project is an intelligent multi-agent AI platform built to transform the way investors discover and evaluate opportunities in emerging cryptocurrencies. The focus is on small and new tokens, where transparency is often low and risks are high. The platform continuously monitors Twitter and other social channels to identify coins that are gaining attention in real time. Using advanced natural language processing and sentiment analysis, it evaluates the tone, intensity, and credibility of these discussions, distinguishing between authentic community-driven growth and artificially coordinated hype.

At the same time, the system integrates real-time on-chain blockchain data to provide a more complete picture of each token’s fundamentals. It examines market capitalization dynamics, tracks how tokens are distributed among holders, and monitors liquidity depth and trading patterns. Suspicious wallet activities such as coordinated buying or selling, sudden spikes in new wallet inflows, or behaviors linked to potential manipulation are flagged automatically.

Behind the scenes, the intelligence is driven by a network of specialized AI agents, each focusing on a distinct responsibility such as sentiment monitoring, blockchain analysis, risk evaluation, and report generation. These agents work together through an orchestration layer, synthesizing their findings into a single coherent investment analysis.

The results are delivered directly to users through a Telegram bot, ensuring real-time accessibility in a format that is both detailed and easy to understand. By combining social intelligence, blockchain analytics, and coordinated agentic AI, the platform functions as an early-warning and discovery system. It empowers investors with transparent, actionable insights, helping them make safer and smarter decisions in the volatile world of emerging cryptocurrencies.

This project was built as a modular multi-agent AI system, where each agent handles a specific layer of crypto intelligence. We started with the ASI Alliance framework for agent orchestration but quickly migrated to 0G Network, which gave us faster throughput, native token integration for analysis credits, and low-cost deployment.

The architecture is composed of four coordinated agents. The Token Discovery Agent scans funding platforms and social media with Firecrawl, sentiment analysis, and bot detection. The Whale Tracking Agent uses custom Graph substreams to monitor the top 100 wallets, detect large transfers, and calculate concentration risk. The Market Data Agent pulls institutional-grade feeds from Pyth Network, computes live valuations, and runs technical indicators. Finally, the Financial Analysis Agent applies game theory modeling and symbolic logic to combine sentiment, whale behavior, and liquidity into composite risk scores.

Each agent is containerized and communicates via lightweight protocols, keeping the system modular and scalable. A Telegram bot interface lets users interact with the platform through simple commands like /analyze <token>, receiving clear and structured analysis reports in real time.

Hackathon constraints pushed us to be resourceful. We batched social data pulls to avoid API limits, spun up agents dynamically to reduce load, and built targeted Graph substreams for faster queries. The 0G integration also enabled a credit-based model where users pay for advanced analyses with tokens, opening the door to governance, staking, and fee-sharing in the future.

Together, these technologies made FinRizz a robust, real-time, and user-friendly AI-powered crypto intelligence platform.