The problem SkyNet solves

In the volatile world of cryptocurrency, staying updated with market trends and executing timely trades can be daunting for both seasoned investors and newcomers. SkyNet addresses these challenges by introducing an innovative solution that simplifies and automates crypto trading through sentiment-driven strategies.

Key Problems Solved:
Trend Identification:
Spotting trending tokens amidst constant market activity is challenging. SkyNet automates this task, using a rag agent to identify promising tokens every 30 minutes, saving users hours of research and analysis.

Community-Driven Insights:
SkyNet fosters community engagement by allowing Telegram group members to discuss identified tokens. This crowdsourced sentiment adds a layer of human insight to market analysis.

Automated Trading Decisions:
Making informed decisions under market pressure is stressful. SkyNet analyzes group sentiment to autonomously execute buy, sell, or hold actions, ensuring rational, data-backed decisions.

Shared Pool Management:
Managing pooled funds requires trust and transparency. SkyNet’s secure wallet infrastructure, powered by Aptos and Okto SDK, ensures that all transactions are safe and verifiable.

Accessibility for All Users:
With Okto SDK enabling wallet creation via Google Auth, SkyNet is user-friendly and accessible, even for crypto novices.

Use Cases:
Effortless Trend Analysis: Traders gain insights into market movements without manual research.
Collaborative Decision-Making: Community input shapes trading strategies, ensuring collective intelligence.
Secure Fund Management: Shared wallets eliminate the need for individual fund handling, reducing risks.
Time Efficiency: Automated actions free users from constant market monitoring.
SkyNet empowers users to navigate the crypto market with ease, leveraging automation and collaboration for smarter, safer trading.

Challenges we ran into

Building SkyNet was both an exciting and challenging endeavor. We encountered several technical and conceptual hurdles during the development process, which tested our skills and resilience.

1. Sentiment Analysis Accuracy:
   A significant challenge was designing an effective sentiment analysis model. Telegram messages often contain slang, emojis, and abbreviations, making it difficult to gauge the sentiment accurately. To overcome this, we incorporated RAG model and fine-tuned it on crypto-specific datasets. The model was iteratively tested and improved by comparing predictions against human-labeled data for better accuracy.

2. Custom JWT using Okto SDK:
   Implementing a custom JWT for wallet generation was tricky due to compatibility issues with different blockchain protocols. This was resolved by thoroughly understanding Okto's architecture and leveraging its detailed documentation to build a seamless integration layer for wallet creation and raw transactions.

3. Scalability of the RAG Agent:
   The RAG (Retrieval-Augmented Generation) agent struggled with processing large datasets for trending token analysis in real time. We optimized it by implementing asynchronous data fetching and processing, reducing latency and enhancing scalability.

4. Managing Pool Funds:
   Ensuring transparency and security in managing shared pool funds was critical. To address this, we used Aptos’ robust transaction infrastructure for clear and auditable operations, reinforcing trust among group members.

5. Real-Time Sentiment Execution:
   Executing trades based on real-time sentiment was a challenge due to fluctuating network latency. We implemented a buffer mechanism that processed sentiments within specific intervals, ensuring consistency in actions without sacrificing real-time efficiency.

Through collaborative troubleshooting, rigorous testing, and leveraging the expertise of our sponsors (Okto and Aptos), we resolved these hurdles.