Crypto X AI的下一轮不仅会留在Memecoin曲目中。

熊市更适合深入研究，了解更多可行的叙述将使我们能够在未来的潮流中脱颖而出。

Crypto X AI's next round will not only stay in the Memecoin track. The bear market is more suitable for in-depth research, and understanding more feasible narratives will enable us to ride the wave to the top in the future.

Crypto X AI的下一轮不仅会留在Memecoin曲目中。熊市更适合深入研究，了解更多可行的叙述将使我们能够在未来的潮流中脱颖而出。

While organizing various reports on the AI track these days, I reflected on the stack released by Coinbase Ventures @cbventures regarding Crypto + AI.

如今，在组织AI轨道上的各种报告时，我在Coinbase Ventures @cbventures发布的有关Crypto + AI的Coinbase Ventures发布的堆栈中进行了反思。

JK @jonathankingvc envisions an ideal scenario for the combination of Crypto and AI: AI Agents interacting across various Crypto infrastructures. Software code (smart contracts) created by AI leads to a surge in Dapp numbers and an enhanced user experience, allowing users to own and govern their own AI large models and profit from them. This can be divided into:

JK @JonathankingVC设想了一个理想的情况，可以在各种加密基础架构上进行互动的加密和AI：AI代理的组合。由AI创建的软件代码（智能合约）导致DAPP数字和增强的用户体验激增，使用户可以拥有并控制自己的AI大型模型并从中获利。这可以分为：

A computing layer primarily based on decentralized computing providers like Aethir.

一个计算层主要基于分散的计算提供商等分散的计算层。

A data layer centered around training datasets to expand AI large models.

一个数据层以培训数据集为中心，以扩展AI大型模型。

A middleware layer composed of various new AI-based infrastructures (training/privacy inference/Agent platforms).

中间件层由各种新的基于AI的基础架构（培训/隐私推论/代理平台）组成。

An application layer.

应用层。

As it stands, there are very few Crypto + AI application layer products that allow retail investors to feel the impact directly, and the experience is poor. The most direct reason is that the foundational layers beneath the application layer have not yet been well established.

就目前而言，很少有加密货币 + AI应用层产品使散户投资者能够直接感受到影响，而且体验很差。最直接的原因是，尚未确定应用层以下的基础层。

Recently, starting from the model itself, FLock.io @flock_io has realized that training large models specifically for the Crypto track can validate its path through on-chain incentives to promote decentralized AI and model training. Although it is working on a larger narrative of decentralized AI model training, validating the feasibility of a standout product in a segmented field during the early stages can help FLock.io accumulate more early supporters, leading to the emergence of FLock.io's Web3 Agent Model.

最近，从模型本身开始，Flock.io @flock_io意识到，专门针对加密训练的训练大型模型可以通过链上激励措施来验证其路径，以促进分散的AI和模型培训。尽管它正在研究分散的AI模型培训的更大叙事，但在早期阶段验证了在分段领域中脱颖而出的产品的可行性可以帮助Flock.io积累更多的早期支持者，从而导致Flock.io的Web3代理模型的出现。

If the application layer's Crypto AI Agent is your smart assistant, then its large model is equivalent to your assistant's brain. Only with deep experience and knowledge can it handle every interaction and execute each instruction correctly.

如果应用程序层的加密AI代理是您的智能助手，则其大型模型等同于您助手的大脑。只有具有深厚的经验和知识，它才能处理每种互动并正确执行每个指令。

The most prominent metric of FLock.io's Web3 AI Agent large model is 75.93% FC precision matching accuracy, which simply means it understands Web3 better. Instructions that other models cannot recognize or that produce nonsensical outputs are addressed by the Web3 AI Agent large model through collaboration with industry partners like IO.net, based on the AI Arena Task 1 collaboration framework. By reducing single data source bias through decentralized training, AI Agents calling upon the Web3 AI Agent Model will be more practical and accurate.

Flock.io的Web3 AI代理大型模型最突出的指标是75.93％FC精确匹配精度，这仅意味着它可以更好地理解Web3。 Web3 AI代理大型模型通过与AI Arena Task 1 Collaboration Framework这样的行业合作伙伴的协作来解决其他模型无法识别或产生荒谬输出的说明。通过通过分散培训来减少单个数据源偏差，呼吁Web3 AI代理模型的AI代理将更加实用和准确。

At this point, we must mention the ecological growth flywheel derived from FLock.io's natural incentive properties as a Crypto product.

在这一点上，我们必须提及从Flock.io的自然激励性能作为加密产品中得出的生态增长飞轮。

By leveraging the basic $FLock incentives, more quality model trainers are encouraged to join, bringing higher quality data and training skills.

通过利用基本的$羊群激励措施，鼓励更多优质的模型培训师加入，带来更高质量的数据和培训技能。

These contributions help build better AI large models;

这些贡献有助于建立更好的AI大型模型；

More powerful AI large models can attract more feature-rich and practical Agents and Dapps to call upon;

更强大的AI大型模型可以吸引更多功能丰富，实用的代理商和Dapps。

Practical DApps and AI Agents, supported by the Web3 model, possess strong market competitiveness to generate more revenue;

由Web3模型支持的实用DAPP和AI代理具有强大的市场竞争力来产生更多的收入；

More calls to the large model by Agents and Dapps will generate more model incentives for trainers;

代理商和DAPP对大型模型的更多电话将为培训师带来更多的模型激励措施；

This new data and usage feedback can further optimize and enhance the AI large model, while the prosperity of the ecosystem will also attract more quality model trainers to join through rewards and recognition.

这些新数据和使用反馈可以进一步优化和增强AI大型模型，而生态系统的繁荣也将吸引更多优质的模型培训师通过奖励和认可加入。

Through this continuously looping positive feedback, the FLock.io ecosystem can achieve sustained growth and self-enhancement. AI Infra is still in its early stages, let alone the Crypto AI field. We look forward to seeing more foundational BUIDLers like FLock.io in the industry. Only by laying a solid foundation for the model can we witness a more stable and flourishing application layer explosion.

通过这种不断循环的积极反馈，Flock.io生态系统可以实现持续的增长和自我增强。 AI Infra仍处于早期阶段，更不用说加密AI领域了。我们期待在行业中看到更多的基本造物主。只有为模型奠定坚实的基础，我们才能目睹更稳定和繁荣的应用层爆炸。