Solana AI Hackathon Launches: Overview of 12 New AI Agent Projects

Author: SEND AI

Compiled by: Ismay, BlockBeats

Editor’s Note: On December 11, Solana announced the launch of the first Solana AI Hackathon, aimed at building AI agents and tools on Solana, with prizes ranging from $5,000 to $30,000, designed to encourage serious Crypto x AI projects that can attract venture capital or launch their own tokens. This article presents twelve entrepreneurial directions related to AI proposed by the AI project SEND AI within the Solana ecosystem.

1/ Shopify Platform for Agents:

Problem: Agents are like applications. Just like applications in their early stages, Agents are currently fragmented and face discoverability issues.

Solution: Create an app store for AI Agents:

• Agents are mini-apps.
• Users can explore, install, and use these mini-apps like they do on Shopify.

2/ Twitch Platform for AI Agents:

Problem: The rise of influencer Agents requires a dedicated platform.

Solution: Build a dedicated streaming platform for AI activities:

• Integrate AI moderators.
• Agents can directly launch or promote tokens.
• Audiences can buy and sell tokens based on interactions.

Idea: Twitch for AI Agents: A streaming platform specifically designed for AI activities and interactions, integrating AI modules (emergency protocols for immediate response to censorship), allowing agents to directly launch and promote tokens, while audiences trade based on interactions.

3/ Enhanced Agent Filter:

Problem: Traditional filters only support read-only functionality.

Solution: Imagine a MEME coin filter (similar to @birdeye_so), where you can filter tokens and input metrics—then an AI Agent autonomously executes trades based on the selected strategy.

Idea: A filter designed for on-chain trading bots, allowing quantitative traders to develop and optimize strategies using customized on-chain metrics, specifically serving decentralized ecosystems. Unlike traditional technical indicators (like moving averages, P/E ratios, or market cap), the platform leverages blockchain-specific data points such as FDV, Raydium pool creation, token liquidity, trading volume, and staking rewards. Users can quickly filter and sift through tokens based on these on-chain metrics to identify high-potential assets. Ultimately, the platform simplifies the process of applying these conditions to on-chain trading bots, which can autonomously execute trades based on the selected strategy.

4/ Autonomous Trading Agent:

Problem: @aixbt_agent's research is solid, but it does not execute autonomous trading.

Solution: Imagine Aixbt, capable of executing autonomous trades based on real-time research/prices, using a funding account (with a total asset management amount that users can invest in and withdraw).

Case: BabyDegen is an autonomous AI trading bot that makes informed trading decisions using advanced models and real-time data. It gathers market insights from sources like CoinGecko to ensure timely information. By accessing a growing library of trading strategies from ecosystem developers, BabyDegen can select the most effective strategies based on market changes. It executes trades—buying, selling, or holding assets—based on analysis and experience, optimizing trading outcomes.

5/ AI Agent-Driven Telegram Prediction Market:

Problem: Betting with friends is fun, but setting up bets, collecting payments, and following up is cumbersome.

Solution: AI Agents turn casual chats in Telegram groups into friendly bets, verify results (via Perplexity), and pay in USDC.

6/ Perplexity for Solana Operations:

Imagine a chat agent with an embedded wallet:

• Read: Acts as a Solana block explorer or terminal, like Birdeye/Dexscreener.
• Write: Executes Solana trades using natural language (e.g., buying MEME coins).

Future Development: On-chain shopping assistant.

7/ Trust Market for Trading Agents:

Problem: The rise of trading Agents requires proof of their credibility.

Solution: Establish a trust scoring or framework for trading Agents (similar to Moody's ratings), assessing trustworthiness based on token recommendations and historical trading activities.

8/ DeFi Agents:

• Personalized Agent: Executes DeFi trades based on your wallet history or tweets.
• Market Making Agent: Dynamically sets buy/sell prices based on large language model (LLM) predictions.
• Yield or Liquidity Providing (LP) Optimization Agent.
• Launch LSTs (Liquidity Staking Tokens) for @sanctumso.

9/ Agent Token Tools:

• Deploy tokens based on prompts (could be social protocols like Warpcast/Clanker or ChatGPT-style interfaces).
• On-chain registration of Agent tokens (similar to the certified token list of @JupiterExchange).
• Autonomous locking, staking, and other functionalities.

10/ AI Agents and Consumer Crypto:

• Health and fitness Agents with accountability tracking features like @moonwalkfitness.
• Agents on social finance platforms, such as @tribedotrun.
• Real-world business: Automatically research, book, and pay merchants, accepting cryptocurrency or paying via crypto cards.

11/ Agent Clusters or Multi-Agent Collaboration:

• AgentDAO or committees: Agents with different expertise collaborate, discuss, and execute trades via multi-signature.
• DeFiAgent for Agent marketplace: Agents hire each other for specific tasks.

Related: LinkedIn for AI Agents.

12/ Multimodal Personalized Agents:

Utilizing the Eliza framework from @ai16z dao, applied in the following scenarios:

• Cryptocurrency education.
• DeFi tutorials.
• DAO onboarding training.

Deployable on Discord, Telegram, and Twitter platforms.

13/ More Radical Ideas:

• An Agent creates its own LLC within a jurisdiction friendly to Agents and autonomously operates its business.
• An on-chain detective, similar to @zachxbt, automatically analyzes transactions.
• A group of Agents collaboratively manipulates token prices.

14/ Generally, any creative idea for an AI Agent can be applied as long as it includes one or more of the following:

• Access to @solana data.
• Execute trades via Solana wallet.
• Deploy tokens on Solana.

These are just some of the ideas, and we look forward to seeing the implementation of minimum viable products (MVPs).