Kolective

Demo

About

The problem it solves Web3 trading is fragmented, opaque, and difficult to navigate. Retail traders struggle with: - Information overload from social media and influencers - Manual execution of trades across volatile markets - Lack of trustworthy guidance and real-time risk management Kolective addresses these challenges by delivering an AI-driven platform that automates trading based on real-time signals from verified Key Opinion Leaders (KOLs), personalized risk profiles, and on-chain data. **Solution Overview** Kolective simplifies the trading experience across decentralized markets by offering: - Personalized Risk Assessment Users undergo a brief questionnaire to determine their investor profile—Conservative, Balanced, or Aggressive. AI then tailors trading strategies accordingly. - Curated KOL Recommendations The platform recommends top-performing KOLs based on credibility, historical trading performance, and alignment with the user’s risk profile. - Automated Trading Execution An AI agent wallet executes buy/sell orders based on KOL signals. Users are not required to monitor markets or manually place trades. - AI-Based Risk Management The system applies dynamic stop-loss and profit-taking thresholds to manage risk and protect capital based on the user's selected profile.

How it's made

Challenges we ran into 1. Integrating Real-Time AI with On-Chain Execution Combining off-chain AI (e.g., tweet parsing, risk profiling) with real-time, on-chain smart contract actions was complex. Ensuring low-latency, reliable syncing between AI decisions and smart contract calls required careful architecture. 2. Parsing Financial Signals from KOL Tweets Extracting accurate trading signals from unstructured, informal tweet content introduced ambiguity. We had to fine-tune NLP pipelines to avoid misinterpreting noise or hype as actionable trades. 3. Balancing Automation with User Control Building a system that automates trades while still giving users full control (manual overrides, portfolio withdrawal, risk adjustment) demanded a well-designed user flow and contract safety mechanisms. 4. Gas Efficiency on Base L2 Although Base is optimized for low fees, batching multiple user actions (auto-sell, rebalancing, stop-losses) into gas-efficient transactions required optimizations in how and when contract functions were called. 5. Risk Profile Calibration Translating qualitative user input (e.g., “I want moderate risk”) into quantitative stop-loss/profit-taking rules and allocation weights required iteration and testing to feel intuitive and effective. 6. Time Constraints for Full Decentralization Due to hackathon limits, some AI functions (like tweet analysis and signal ranking) were kept off-chain for performance reasons. Bringing this logic fully on-chain remains a future goal using verifiable computation.