Original author: SEND AI
Original translation: Ismay, BlockBeats
Editors Note: On December 11, Solana announced the launch of the first Solana AI Hackathon, which aims to build AI agents and tools on Solana. The prizes range from $5,000 to $30,000, and are intended to encourage serious Crypto x AI projects that can attract venture capital or launch their own tokens. This article is about twelve entrepreneurial directions for AI proposed by SEND AI, an AI project in the Solana ecosystem.
1/ Agent’s Shopify platform:
Problem: Agents are like applications. Just like the early days of applications, Agents are currently fragmented and face discoverability issues.
Solution: Create an app store for AI Agents:
–– Agent is a mini application.
–– Users can explore, install, and use these mini apps just like they do with Shopify.
2/ AI Agent’s Twitch platform:
Problem: The rise of influencers requires a dedicated platform.
Solution: Create a dedicated streaming platform for AI activities:
–– Integrated AI Moderator
–– Agents can directly launch or promote tokens
–– Viewers can directly buy and sell tokens based on their interactions
Idea: Twitch for AI Agent: A streaming platform built specifically for AI activities and interactions, with an integrated AI module (an emergency protocol for immediate response to censorship), where agents can directly launch and promote tokens, and viewers can buy and sell based on the interaction.
3/ Enhanced Agent Filter:
Issue: Classic filters support only read-only functionality.
Solution: Imagine a MEME coin screener (similar to @birdeye_so) where you can filter tokens and enter metrics - then an AI Agent autonomously executes trades based on the selected strategy.
Idea: A filter designed for on-chain trading robots that allows quantitative traders to develop and optimize strategies using tailored on-chain indicators, specifically for the decentralized ecosystem. Unlike traditional technical indicators such as moving averages, P/E ratios, or market capitalization, the platform utilizes blockchain-specific data points such as FDV, Raydium pool creation, token liquidity, trading volume, and staking rewards. Users can quickly screen and filter tokens based on these on-chain indicators to identify high-potential assets. Ultimately, the platform simplifies the process of applying these conditions to on-chain trading robots, which can autonomously execute trades based on the selected strategy.
4/ Autonomous Trading Agent:
Question: @aixbt_agents research is very solid, but it does not perform autonomous trading.
Solution: Imagine Aixbt, with the ability to execute autonomous trades based on real-time research/prices, using one funding account (with AUM that users can invest and withdraw).
Case: BabyDegen is an autonomous AI trading robot that uses advanced models and real-time data to make smart trading decisions. It collects market insights from sources like Coingecko to ensure the timeliness of information. By accessing a growing library of trading strategies from ecosystem developers, BabyDegen is able to select the most effective strategy based on market changes. It executes trades based on analysis and experience—buying, selling, or holding assets—to optimize trading results.
5/ Telegram prediction market driven by AI Agent:
Problem: Placing bets with friends is fun, but setting up bets, collecting payments, and following up can be tedious.
Solution: AI Agent turns small talk in Telegram groups into friendly bets, verifies the results (via Perplexity), and pays out USDC.
6/ Perplexity for Solana Operations:
Imagine a chat agent with a wallet embedded in it:
– Read: Acts as a proxy for Solana block explorers or terminals, such as Birdeye/Dexscreener.
– Write: Execute Solana transactions (e.g. buy MEME coins) using natural language.
Future development: on-chain shopping assistant.
7/ Trading Agent Trust Market:
Problem: The rise of trading agents needs to prove their credibility.
Solution: Establish a trust score or framework for trading agents (similar to Moodys ratings) to assess trust based on token recommendations and historical trading activities.
8/ DeFi Agent:
–– Personalized Agent: Execute DeFi transactions for you based on your wallet history or tweets.
–– Market Maker Agent: Dynamically sets buy/sell prices based on Large Language Model (LLM) predictions.
–– Yield or liquidity provider (LP) optimization agent.
–– Launching @sanctumso’s LSTs (Liquidity Proof Tokens).
9/ Agent Token Tools:
–– Deploy tokens based on prompts (could be a social protocol like Warpcast/Clanker or a ChatGPT style interface).
–– On-chain registry of Agent tokens (similar to @JupiterExchange’s list of authentication tokens).
–– Functions such as autonomous lock-up and pledge.
10/ AI Agent and consumer encryption:
–– Health and fitness Agent, with @moonwalkfitness type accountability tracking capabilities.
–– Agents on social finance platforms, such as @tribedotrun.
–– Real World Commerce: Automatically research, book, and pay for merchants that accept cryptocurrencies or pay via crypto cards.
11/ Agent cluster or multi-agent collaboration:
–– AgentDAO or committee: Agents with different expertise collaborate, discuss, and execute transactions through multi-signature.
–– DeFiAgent to Agent Marketplace: Agents hire each other for specific tasks.
Related: AI Agent on LinkedIn.
12/ Multimodal Personalized Agent:
Using @ai16z daos Eliza framework, applied to the following scenarios:
–– Cryptocurrency Education
–– DeFi Tutorial
–– DAO onboarding training
Deployable on Discord, Telegram and Twitter platforms.
13/ More radical ideas:
–– An agent creates his or her own LLC in an agent-friendly jurisdiction and operates his or her own business autonomously.
–– An on-chain detective, similar to @zachxbt, automatically analyzing transactions.
–– A group of agents work together to manipulate the token price.
14/ In general, any AI Agent idea can be applied as long as it contains one or more of the following:
–– Access @solana data
–– Execute transactions through Solana Wallet
–– Deploying Tokens on Solana
These are just some of the ideas we’re looking forward to seeing implemented into minimum viable products (MVPs).
