正在加载视频...
视频加载失败
Decentralized Autonomous AI Agents: The Future of Smart Contracts Truly gamifying DeFi with AI… omg buzzwords, ban+report?! Not quite. We can make strategies, gamified entries-exits, conditions for positions, etc. All are accessible via simple text. Intelligent DeFi made real. I’m excited to share my research on AI-driven autonomous agents!... show more
10 条评论

🔹 Scaling Programmable Money Developing smart contracts isn’t easy—it requires a solid computer science background. Developers today are constrained by the EVM’s size limits, which restrict the complexity of autonomous strategies. My work explores a path beyond these limits.

🔹 What Are DA-AI Agents? Deterministic Large Language Models (LLMs) functioning as @eigenlayer AVS. They monitor on-chain events and make decisions accordingly. Model parameters and prompts are stored in smart contracts, ensuring the same output for identical inputs, making their actions fully verifiable. Any malicious behavior? Slashable! For example, you can set up a strategy to "Buy ETH if Trump wins." Review it twice; this is the final version of the code that you will deploy on-chain to launch your bot.

🔹 Real-World Oracles Lower gas costs and improved Data Availability (DA) enable us to bring more data onto the blockchain. Currently, there is little incentive to deliver textual data to the blockchain. I believe that DA-AI is a game changer. Just imagine oracles streaming @VitalikButerin’s tweets directly on-chain. You could easily build a bot that generates meme coins triggered by each tweet. Does that sound futuristic? It’s already been implemented in a prototype.

🔹 Anyone Can Be a Smart Contract Developer Getting started with a DA-AI agent is super easy! There's no need to learn coding languages like Solidity or Rust—just express your commands in plain English or any language the model understands. Let's stop talking about AI x Crypto and start building real products!

Petty cool demo, Can't wait till AI dumps on me another shitcoin 🤣. Couple of comments: 1. There are no "fair systems" to get an edge you need an informational advantage, this is the sole reason, bots, HFTs, CEX-es exist. Building a public source of info in this regards achieves nothing. 2. Have you considered inflation and informational aspects of launching shitcoins with AI agents? The Value of any single coin will tend to zero eventually and we will see AI agents flipping those coins constantly. The hardest part will be finding that retail liquidity to dump on. AI fighting AI and scamming retail. Not the exact future I'd love to see! 3. One prominent approach for AI in DeFi (at lest for me) can be described simply as: yield optimising agents. Probably, should add some RL into the mix to learn from the environment. I'd love to see AI as an cross-chain universal solver, which would provide me best investment opportunities, portfolio breakdowns, and trade routs based on the info it monitors from the outside world. 4. Why do you call LLMs deterministic? In a nutshell they output a probability distribution, so they are probabilistic at best. At worst, no one really knows how LLMs are able to "reason". There have been numerous studies about that, but we're not anywhere close to the answer.

Thanks for the great questions: 1. Fairness and access is programmable onchain. You can set up a cooldown period, priority rules via auctions, etc. Until they're clear and transparent, the system is fair.

this is great work, see you soon! 🇹🇭👋

Completely aligned! While traditional smart contracts are powerful, they lack adaptability. And AI integration could change that. We at GenLayer are enabling Intelligent Contracts that can leverage LLMs to access real-time data and make complex decisions.

Did u already present this? Would love to learn more

Right there with you, Mikko! Devcon week in Bangkok is packed with amazing events, and we’re wrapping it up with an afterparty for the community. Let’s connect! 🎶✨
