Pond 是一家构建去中心化 AI 模型层的加密初创公司，筹集了 750 万美元的种子资金

Pond 成立于 2023 年，最初构建了一个由链上数据驱动的用户搜索引擎，使用户能够探索区块链连接。

Crypto startup Pond, which is building a decentralized AI model layer, has raised $7.5 million in a seed funding round led by Archetype.

加密初创公司 Pond 正在构建去中心化人工智能模型层，在由 Archetype 领投的种子轮融资中筹集了 750 万美元。

Other participants in the round included Cyber Fund, Delphi Ventures, Coinbase Ventures, Near Foundation and several angel investors, such as Illia Polosukhin, co-founder of Near Protocol; Chris Yin, CEO and co-founder of Plume Network; Cynthia Wu, founding partner and COO of Matrixport; and Tesa Ho, head of market research at Flashbots, Pond announced Thursday.

此轮融资的其他参与者包括 Cyber​​ Fund、Delphi Ventures、Coinbase Ventures、Near Foundation 和几位天使投资人，例如 Near Protocol 联合创始人 Illia Polosukhin； Chris Yin，Plume Network 首席执行官兼联合创始人； Cynthia Wu，Matrixport 创始合伙人兼首席运营官； Pond 周四宣布，Flashbots 市场研究主管 Tesa Ho 表示。

Pond began raising money for this round in April and closed it by July, co-founder and CEO Dylan Zhang told The Block. The round was structured as a simple agreement for future equity (SAFE) with token warrants, said Zhang, who declined to disclose the post-round valuation.

Pond 联合创始人兼首席执行官 Dylan Zhang 告诉 The Block，该轮融资于 4 月份开始筹集，并于 7 月份关闭。张表示，本轮融资的结构是一份带有代币认股权证的简单未来股权协议（SAFE），但他拒绝透露本轮融资后的估值。

What is Pond?

什么是池塘？

Founded in 2023, Pond initially built a user search engine powered by on-chain data, enabling users to explore blockchain connections. For example, if someone wanted to connect with Ethereum co-founder Vitalik Buterin, they could enter his ENS (vitalik.eth) and view who is connected to him in their network. "We achieved this through a proprietary graph algorithm designed specifically for blockchain data, which calculates the strength of these connections," said Zhang.

Pond 成立于 2023 年，最初构建了一个由链上数据驱动的用户搜索引擎，使用户能够探索区块链连接。例如，如果有人想与以太坊联合创始人 Vitalik Buterin 联系，他们可以输入他的 ENS (vitalik.eth) 并查看谁在他们的网络中与他连接。 “我们通过专门为区块链数据设计的专有图形算法实现了这一目标，该算法计算这些连接的强度，”张说。

Over time, Pond saw the potential of a unified graph network across all types of on-chain data — social, financial and more — and expanded its focus to build crypto-native AI models for broader applications. "In the entire industry of AI, models are the engine that generates productivity," Pond co-founder and CTO Bill Shi told The Block. "By learning from and understanding data, models extract value from data and make predictions. Consider the current AI wave: It was unlocked by models like GPT and Llama."

随着时间的推移，Pond 看到了跨所有类型的链上数据（社交、金融等）的统一图网络的潜力，并将其重点扩展到为更广泛的应用构建加密原生人工智能模型。 Pond 联合创始人兼首席技术官 Bill Shi 告诉 The Block：“在整个人工智能行业中，模型是产生生产力的引擎。” “通过学习和理解数据，模型从数据中提取价值并做出预测。想想当前的人工智能浪潮：它是由 GPT 和 Llama 等模型解锁的。”

Pond is now creating a "complete model ecosystem," encompassing data, model computation, training and inference. Shi said that Pond not only builds its own models but also supports others in model creation and commercialization.

Pond 现在正在创建一个“完整的模型生态系统”，涵盖数据、模型计算、训练和推理。施说，Pond不仅建立自己的模型，还支持其他人的模型创建和商业化。

User perspective

用户视角

While web2 AI models like ChatGPT assist with tasks from text generation to recommendations, Shi said Pond's web3 AI models will power crypto-specific use cases driven by on-chain data.

虽然像 ChatGPT 这样的 web2 AI 模型可以帮助完成从文本生成到推荐等任务，但 Shi 表示，Pond 的 web3 AI 模型将为由链上数据驱动的加密特定用例提供支持。

"On-chain data is extremely messy and vast in quantity, making it incomprehensible to the human mind. By leveraging AI capabilities, we transform what is beyond human understanding into comprehensible information, helping users make better use of on-chain data," he said.

“链上数据极其混乱、数量庞大，人类思维难以理解。通过人工智能能力，我们将人类无法理解的信息转化为可理解的信息，帮助用户更好地利用链上数据。”说。

Currently, Pond supports applications in security, recommendations and DeFi.

目前，Pond 支持安全、推荐和 DeFi 领域的应用。

In security, Pond has developed a graph-based model for predicting malicious addresses, achieving an accuracy of 0.936 and a precision of 0.935. This model will be available on Pond's platform and will be used by "an industry-leading security company," according to Shi.

在安全方面，Pond开发了基于图的模型来预测恶意地址，准确率达到0.936，精度达到0.935。 Shi 表示，该模型将在 Pond 平台上提供，并将由“行业领先的安全公司”使用。

In recommendations, Pond offers models that suggest NFTs to users on Zora and DeFi protocols on Gearbox.

在推荐方面，Pond 提供了向 Zora 上的 NFT 和 Gearbox 上的 DeFi 协议的用户推荐 NFT 的模型。

In DeFi, Pond has developed a dynamic fee model that addresses a key issue for liquidity providers: determining optimal transaction fees, Shi said. "When an intended trade is predicted to potentially lead to losses for liquidity providers, the model increases the transaction fee to compensate for the possible loss. Conversely, when a trade is predicted to be beneficial to liquidity providers, the model can lower the transaction fee to encourage more of these trades," he explained.

Shi 表示，在 DeFi 领域，Pond 开发了一种动态费用模型，解决了流动性提供者的一个关键问题：确定最佳交易费用。 “当预计一项交易可能会给流动性提供者带来损失时，该模型会增加交易费用以补偿可能的损失。相反，当预计一项交易对流动性提供者有利时，该模型可以降低交易费用以鼓励更多此类交易，”他解释道。

Pond plans to expand its crypto-AI use cases to include DeFi risk management, insider trading detection and personalized recommendations. Currently employing 11 people, Pond is actively hiring for roles including chief operating officer, product manager and ecosystem growth manager, said Zhang.

Pond 计划扩展其加密人工智能用例，包括 DeFi 风险管理、内幕交易检测和个性化建议。张说，旁氏目前拥有 11 名员工，正在积极招聘首席运营官、产品经理和生态系统增长经理等职位。

The Funding newsletter: Stay updated on the latest crypto funding news and trends with my bimonthly newsletter, The Funding. It's free. Sign up here!

融资通讯：通过我的双月通讯“The Funding”了解最新的加密融资新闻和趋势。它是免费的。在这里注册！