Agent-Fi on AO:融合AI代理的金融范式

Odaily星球日报Published on 2024-07-07Last updated on 2024-07-07

Abstract

本文主要介绍了 AO 上融合 AI 代理的金融范式:Agent-Fi。

想象在未来的世界中,AI 代理智能体与人类形成一种数字化的伴随/共生关系,自主代理(Autonomous Agent)可以根据用户提出的自然语言需求,在对话中明确意图,自动拆解任务并实现预期结果。

AO 建立了一个基于 Actor 的异步并行网络,不对合约全部计算过程共识,而是通过仅对交易顺序共识,乐观默认固定交易顺序在虚拟机中运行结果一致。这一选择允许了 AO 网络的计算进行大规模扩容,直至支持任意类型的计算。AR 网络被作为交易顺序共识的达成层,与交易结果状态的存储层。

与当前其他主流区块链项目大多作为单体区块链且只从底层支持原生状态机的智能合约相比,AO 的基础设施兼容能支持更复杂的计算能力,这就包括了 AI 模型的运行。

AO 的计算单元(Compute Unit)在最近的 WASM 虚拟机的更新之后,已经能够访问 16 GB 内存,这意味我们能够在 AO 上下载和执行 16 GB 的模型。16 GB 已经足够运行大语言模型计算,比如 Llama 3 未量化版本的 Falcon 系列以及许多其他模型。

同时,AO 使用 WeaveDrive,让用户可以像访问本地硬盘一样访问 AO 内的 Arweave 数据,并且兼容不同类型的虚拟机高度异构的进程在一个共享环境中交互,这代表我们享有更多的数据源和组合可能性。这也意味着在未来构建应用程序时,用户上传数据到 Arweave 的动机增加,因为这些数据也可以在 AO 程序中使用。AO 开发团队在测试大型语言模型在 AO+AR 系统中运行时,已经大约上传了价值 1000 美元的模型数据到网络上,但这仅仅是开始。

AO 的系统设计让实施融合 AI 代理的智能合约成为可能。通过在 AO 中编程,我们创造 AI 代理在市场中做出智能决策,代理可能相互对抗,也可能代表人类对抗人类。“当我们审视全球金融体系时,纳斯达克大约 83% 的交易是由机器人执行的。” 当下的量化交易是 AI 代理交易的前身,而未来设计并选取机器学习模型,执行自动化交易的过程会更轻易地被 AI“开盒”并自动化。

过去几年中 DeFi 的发展使得在链上执行各种金融操作可无需信任中心化的实体,如借贷,交易代币或是衍生品。但我们真正谈论市场时,不仅仅是这些操作的可靠性,事实上,可靠的执行各种操作仅仅是基础。决定一个市场是否有活力的核心因素依旧是资本的流动,是决定买卖、借贷或参与各种金融游戏的人。在当下,如果你想参与加密货币投资,而不想自己做所有研究和参与,你必须找到一个可靠的基金,信任他们管理你的资金并放权给基金成员去执行智能决策。但伴随着 AO 应用的发展,我们或许就能扩宽市场的智能决策部分,在网络中筛选信息,加工数据,组合策略,融合 AI 代理的智慧在网络中实时决策,创建非常丰富的去中心化自主代理金融系统。

当下已经有一些项目开始实现这一愿景,我们将介绍 Autonomous Finance(以下简称 AF)、Dexi 与 Outcome,其中,AF 的成果最为瞩目。

Autonomous Finance

AF 专注于在 AO 上研究和开发结合 AI 的金融应用,通过在 AO 链上构建 AI 模型和数据驱动的金融决策,AF 做出了将智能决策层上链的尝试。主要业务有 3 个部分,分别是核心设施(Core Infrastructure)、智能代理金融(AgentFi)和内容金融(ContentFi)。

Agent-Fi on AO:融合AI代理的金融范式

核心设施包含了去中心化交易所(DEX)、借贷、衍生品以及合成资产等协议。

AgentFi 主要指通过可组合的半自主和完全自主代理来实现交易策略的执行。与其他依赖链下程序进行信号处理和逻辑处理的自主代理框架不同,AF 提供的自主代理使用链上数据流进行自我学习,在 AO 生态系统内的各个流动性池与金融基础上执行投资策略。这些代理可以自主运行,无需链下信号或人工干预。

典型的自主代理包括:

  • 美元成本平均法 (DCA) 资产管理代理

  • 自平衡自主指数基金

  • 具有定制风险策略的自主对冲基金

  • 收益聚合代理

  • 链上预测代理

  • 高频交易代理

其中 DCA 代理作为基础代理,在其他更复杂的代理执行逻辑时常被调用,所以作为一个频繁使用的可组合代理模块有许多可定制参数供用户根据自己的需求调节,比如特定价格区间内的触发交易,固定间隔交易时间长度的调节和基于资产价格加权交易(如价格较低时买入更多),还有数据驱动的止盈和利润再投资信号。

DCA 代理应用围绕两个关键的 AO 流程构建:

  • 带有 Cron(基于时间的任务管理系统,常用于定时触发任务执行)触发的代理进程:主要负责进行用户发起与自动定时的 DCA 交易,记录管理的资金并及时更新后端的 AO 进程

  • 后端的 AO 进程:管理与用户名下相关的代理应用并跟踪记录每个代理的历史交易

下图说明了 DCA 代理的设计构架与交互组件

Agent-Fi on AO:融合AI代理的金融范式

对于使用前端的用户来说,DCA 代理的前端基于 DEXI 构建,用户可以通过在 DEXI 网站上连接 AO Connect 钱包来进行 DCA 代理的设置。其中 DEXI 访问有关可用 AMM 池的信息并获取最新价格,DCA 代理负责执行具体的交易逻辑,后端 AO 进程检索与用户相关的所有代理。

Agent-Fi on AO:融合AI代理的金融范式

内容金融是一个框架,用于将存储在 Arweave 永久网络上的数据归因并货币化为 AO 流程的可组合资产。AF 正在构建应用程序,允许数据贡献者或内容基金向 permaweb 贡献例如历史和实时市场情报的数据。而这些内容将作为自主代理和机器学习的链上信号。比如,自主代理会根据社交媒体情绪和历史数据创造新市场。一些示例:

  • 将数据信号货币化

  • 内容驱动的财务代理

  • 基于订阅的数据推荐代理

  • 有影响力的人为自主财务策略贡献数据

  • 数据贡献相关的 DAO 和内容基金聚合各种数据源,以提供动态链上信号

目前,AF 已上线两个主要的产品,分别是 AO Link 和 Data OS。

AO Link  是 AO 网络的消息浏览器,提供与传统区块链系统中的区块浏览器类似的功能。它包括消息计算功能、消息链接的图形可视化(清晰易懂)、实时消息流(最新信息)以及链接消息列表(便于组织导航)。用户还可以查看其代币余额和消息收件箱。此工具提供了一种专业而高效的方式来与 ao 网络的结构和活动进行交互和分析。

Data OS 是在 AO Network 上开发的 ContentFI 协议,它采用自主 AI 代理来获取内容、再生成内容衍生品。通过这种创新方法,DataOS 不仅增强了内容的相关性和可访问性,还为内容创建者建立了奖励机制。目前我们可以在https://stats.dataos.so/ 中查看 AO 网络上的各类数据,观察网络活跃度,与内容相关的各种数据暂时没有展示。

Dexi

Dexi 是普通用户在 AO 中使用代理参与 Agent Fi 至关重要的交互界面,它同时也是 AO 网络上的由代理实现的一个应用程序,可以自主识别、收集和汇总 AO 网络中各种事件的各种财务数据(相当于 AO 上的 Dexscrenner)。这些数据涵盖资产价格、代币交换、流动性波动以及代币资产特征(如智能合约详细信息)。Dexi 主要服务于两类用户:通过 Web 终端访问平台的终端用户和和通过发送消息与 Dexi 交互以利用收集的数据的 AO 应用(可理解为 Bot/Agent)。作为核心基础设施,Dexi 主要提供的服务是数据订阅服务,AO 网络上的进程可以付费订阅 Dexi 的数据流,并立即收到价格调整等更新的警报。

Outcome

Outcome 是@puente_ai 团队构建的一个预测市场(prediction market),受到了@fwdresearch@aoTheVentures@aoComputerClub的支持。Outcome 为用户提供一个可对各种事件进行下注的平台,目前市场中的预测主题涵盖科技,迷因(Memes),商业,游戏,DeFi 与 AO。项目声称未来用户可以通过构建依靠现实数据,基于大语言模型的自主代理来进行预测市场的自动下注。

AO 上的 AgentFi 为我们提供了一个新的视角,探索未来在区块链上直接进行 AI 模型部署并使用各种 AI 代理来执行自动化交易。传统单体区块链的限制被 AO+AR 的设计用新颖的底层创新打破,我们期待看到更多 AO 上的应用和结合 AI 代理实现金融策略的案例。

参考

https://www.theblockbeats.info/news/53865

https://permadao.com/permadao/AI-on-AO-AO-AI-224ba15c840a4309972fec5350d9ed90

https://www.communitylabs.com/blog/ao-in-ai-key-highlights?utm_source=Blog&utm_medium=X&utm_campaign=AI+on+AO&utm_id=Community+Labs

https://www.autonomous.finance/research/en-US

Trending Cryptos

Related Reads

Bitcoin at 59,000 Is Not the Bottom, One Last Drop Needed! Chain Data and Liquidity Analysis: Where is BTC's True Bottom?

Based on analysis by trader Mr. Beggar, Bitcoin's (BTC) recent low of $59k is likely not the final cycle bottom. He argues that while a bottom is near, a final downward movement is still probable to target liquidity below that level, making a deeper low healthier for a sustainable reversal. Mr. Beggar's framework combines on-chain data for long-term cycles and liquidity-based technical analysis for shorter-term trades. His "four deep bear buying models" include Cointime Price (market cost weighted by coin holding time) and AVIV (an enhanced MVRV indicator), which currently suggest prices are nearing cyclical bottom zones. While a PSIP (Percent Supply in Profit) signal has flashed below 50%, it alone is not considered definitive; typically, the first signal is not the final bottom. He presents three potential scenarios for the current market: 1) a direct drop from here, 2) an upward liquidity sweep (stop hunt) of the recent high near $67.3k before declining, and 3) a direct reversal without new lows. He heavily discounts the third scenario due to significant un-swept liquidity in the $59k-$62.3k range, suggesting the market must revisit these levels. Mr. Beggar shares that he used on-chain signals to identify potential cycle tops in late 2024/early 2025 and later established low-leverage BTC-denominated short positions. He emphasizes the importance of risk management and staying within one's expertise ("strike zone"), warning against investing in assets like AI/semiconductor stocks simply because they are rising.

marsbit11m ago

Bitcoin at 59,000 Is Not the Bottom, One Last Drop Needed! Chain Data and Liquidity Analysis: Where is BTC's True Bottom?

marsbit11m ago

From Signal Monitoring to Strategy Copy Trading: How PPP Lowers the Barrier to Trading on Polymarket?

From Signal Monitoring to Strategy Copy Trading: How PPP Lowers the Barrier to Polymarket Trading The surge in trading demand on prediction markets like Polymarket, especially during events like the World Cup, exposes a common challenge for novice users: emotional and impulsive trading due to a lack of stable strategies and reliable signals. Prediction Position Platform (PPP) addresses this by serving as a Telegram-based tool for strategy discovery and automated copy-trading on Polymarket. PPP offers a suite of features through a subscription model. Key functionalities include 24/7 market signal monitoring (tracking smart money movements and rapid probability shifts), an "AI Address Analysis" tool to evaluate trader performance metrics, and specialized sections like a "World Cup Zone" for quick access to related markets. Its core value lies in two curated lists: the "Strategy Square," which identifies addresses suitable for long-term tracking based on comprehensive metrics like returns, win rate, and drawdowns, and the "Trading Leaderboard," highlighting recently outperforming addresses for short-term opportunities. Users can manually analyze any address or set up automated copy-trading with customizable parameters like investment amount and stop-loss. After initiating copy-trades, users can manage all positions from a unified dashboard, adjusting parameters or stopping follows as needed, and review historical performance data. Crucially, PPP employs a non-custodial wallet model, meaning user funds remain in their own self-custodied wallets, enhancing security and trust. In summary, PPP aims to reduce the learning curve and trial-and-error cost for Polymarket users by aggregating signals, curating and analyzing profitable traders, and facilitating automated, yet manageable, copy-trading execution.

Odaily星球日报11m ago

From Signal Monitoring to Strategy Copy Trading: How PPP Lowers the Barrier to Trading on Polymarket?

Odaily星球日报11m ago

From the White-Haired Stock God to the Billion-Dollar Fund Titan: The Smart People Shorting NVIDIA Are Getting Rich Using the Same Framework

From "white-haired stock god" to billionaire fund manager, those profiting from shorting NVIDIA share a common framework. The article analyzes the critical bottlenecks in the AI hardware supply chain, which have become key investment focal points. The core argument is that the real constraint on the AI boom isn't software or algorithms, but fundamental physical infrastructure. The piece dissects nine major bottlenecks, organized around the lifecycle of an AI accelerator circuit board. *Before the Board*: The pre-manufacturing stage faces constraints in EDA tools, new materials (like GaN, SiC, InP) replacing silicon, and the critical, non-renewable supply of helium for semiconductor fabrication. *On the Board*: The primary bottlenecks are High-Bandwidth Memory (HBM), essential for unleashing GPU power, and advanced packaging (e.g., CoWoS), required to integrate components. Both are in severe shortage. *Between Boards*: Chip-to-chip communication is hitting limits with copper, pushing photonics and optical interconnects (CPO) as the next-gen solution, with NVIDIA heavily investing in this area. *Around the Board*: Power delivery requires new materials (GaN/SiC) for efficient voltage conversion from 48V to sub-1V. High-density AI racks (120kW+) are forcing a shift from air to liquid cooling as the standard. *Beyond the Board*: The ultimate bottleneck is electricity. AI data centers consume power equivalent to mid-sized cities, and grid expansion lags far behind demand, causing project delays and a scramble for power contracts. Prominent investors like Leopold and "white-haired stock god" are heavily betting on these infrastructure bottlenecks. Leopold's fund, for instance, holds no NVIDIA stock but uses massive put options to short the semiconductor sector while going long on power and physical infrastructure. His thesis is that while chip competition may eventually erode margins, the scarcity of foundational elements like electricity is more persistent. The framework's validity is tied to the supply-demand gap. Major new capacity in HBM and photonics is scheduled for 2027-2028, but demand continues to outpace it. Experts like Intel's CEO suggest no relief before 2028. However, the article warns of a potential reversal around 2028-2029 if AI capex slows and new capacity floods the market, turning scarcity into oversupply. Until then, the imbalance persists.

链捕手43m ago

From the White-Haired Stock God to the Billion-Dollar Fund Titan: The Smart People Shorting NVIDIA Are Getting Rich Using the Same Framework

链捕手43m ago

Trading

Spot
Futures

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片