From trading to auditing, take a quick look at 10 Cookie DeFAI hackathon projects worth following

Reprint: Mars Financial, Daisy

Cookie DeFAI Hackathon has ended, showcasing the transformative potential of AI in financial innovation. This hackathon received submissions from over 80 projects across 6 competitive verticals, with a total of 18 projects winning awards. These projects demonstrated excellence in various aspects:

Cross-chain interoperability design

New AI applications in DeFi

Performance optimization solution

Unfortunately, tokenized projects have been disqualified, even though these projects have more mature products, rather than just staying at the conceptual level. Hopefully, the next Cookie DeFAI hackathon will involve more types of participants, including those with tokens.

To cut a long story short, from the list of winners, it can be seen that most of the winning projects are brand new. This article will introduce 10 interesting projects that appeared in the hackathon—starting from the disqualified projects, and then introducing the winning projects.

AlphaSwarm

AlphaSwarm is a DeFAI autonomous trading agent toolkit supported by Python, which can be used to build AiFi agents driven by LLM. AlphaSwarm provides building blocks for quickly and efficiently creating complex, autonomous financial agents. With AlphaSwarm, users can:

Obtain data from multiple sources

Develop the AiFi agent using modular strategies and complex planning driven by LLM

Integrated with DEX

Interact with users through multiple channels

Edwin

Edwin is a TypeScript-driven infrastructure for AI agents to interact with Defi protocols, supporting the creation of DeFAI agents (LP, lending, cross-chain interactions).

aiymx is the first real-time LP agent using Edwin infrastructure.

Homo Memetus

Homo Memetus is a DeFAI ecosystem that tokenizes proxy trading strategies, allowing them to compete with each other (or with humans) to identify profitable strategies (through natural selection).

The vision/roadmap of this project is very interesting, but the project is still in a very early stage.

Anti-Rug Agent

Anti-Rug Agent is a comprehensive security tool for Solana, which can analyze liquidity, wallets, internal personnel, bundled packages, and GitHub through RepoAnalyzer.

AI Security Score of Anti-Rug Agent is one of the best tools for due diligence.

BeforeAI (formerly known as CertaiK)

Before AI (on Virtuals) can achieve large-scale democratization of smart contract audits, allowing AI-driven audits to be completed within 30 seconds.

The team recently launched the AI Agent security score and LARP Analysis new features.

Honorary Nomination Project

JailbreakMe (Token)

According to the official introduction, JailbreakMe is the first AI security platform launched fairly, and users can obtain rewards by cracking AI agents. Jailbreak previously won third place in the Solana AI Hackathon.

AgentXYZ

AgentXYZ is a trading terminal with social and on-chain insights. Recently launched an AI rating system, providing buy/sell recommendations.

The following are some winning projects (no tokens)

Kudo

Kudo has introduced a contract framework, which implements complex financial use cases between agents through contract NFT (cNFT), such as uncollateralized loans, employment agreements, etc.

Kudo’s framework enables autonomous AI agents to establish contracts by minting NFTs representing their commitments. This unlocks some use cases such as allowing AI agents to obtain unsecured loans, establish political alliances to achieve common goals, etc., by setting prompt verification and TEE proof to ensure the integrity of commitments.

Egg AI

Egg AI is a copy trading agent with AI research, market analysis, and execution functions. It combines the SocialFi layer + AI DYOR and execution layer.

Two modes:

Tren Egg (Fully Automatic Copy Trading)

Chad Egg (customizable control)

Aevia

Aevia is the first agent to help users preserve their legacy. When a user passes away, information, cryptocurrencies, NFTs, and all content can be programmatically passed on to the user’s loved ones. The team also has an agent that regularly checks if the user is safe.