One article takes stock of the 10 major integrated development directions of Crypto AI: interaction between agents, content marketing and data market, etc.

Original title: "Crypto x AI: 10 Categories We're Watching in 2025"

Author: Archetype

Compiled by: Shenchao TechFlow

1. Agent-to-Agent Interaction

Due to its natural transparency and composability, blockchain has become an ideal platform to achieve seamless interaction between intelligent agents. In this interaction, agents developed by different institutions for different purposes can collaborate to complete tasks. There are already some exciting attempts, such as transferring money between agents and issuing tokens together , etc. We expect the interaction between agents to be further expanded: on the one hand, create new application scenarios, such as new social platforms driven by agents ; on the other hand, optimize existing enterprise workflows , such as platform authentication, micropayments, and cross-platform work. Stream integration, etc., thus simplifying today’s complex and cumbersome operational processes. - Danny , Katie , Aadharsh , Dmitriy

aethernet and clanker jointly issue Token on Warpcast

2. Decentralized Agentic Organizations

Large-scale multi-agent collaboration is another exciting research direction. How do multi-agent systems collaborate to complete tasks, solve problems, and even manage protocols and systems? In the article "The Promise and Challenges of Crypto + AI Applications" in early 2024, Vitalik proposed the idea of ​​using AI agents for prediction markets and rulings. He believes that in large-scale applications, multi-agent systems have huge potential for "truth" discovery and autonomous governance. We look forward to seeing how the capabilities of this multi-agent system can be further explored, and how “swarm intelligence” can reveal more possibilities in experiments.

In addition, the collaboration between intelligent agents and humans is also a direction worth exploring. For example, how communities interact around intelligent agents, or how intelligent agents organize humans to complete collective actions. We hope to see more experiments with agents targeting large-scale human collaboration. Of course, this requires some kind of verification mechanism, especially if the task is completed off-chain. But this exploration may lead to some unexpected and wonderful results. -Katie , Dmitriy , Ash

3. Agent-driven multimedia entertainment (Agentic Multimedia

Entertainment)

The concept of digital avatars has been around for many years. For example, Hatsune Miku (2007) has held sold-out concerts in 20,000-seat venues; Lil Miquela (2016) has more than 2 million followers on Instagram. Recent examples include AI virtual streamer Neuro-sama (2022), whose subscriber base on Twitch has exceeded 600,000; and anonymous Kpop boy band PLAVE (2023), whose views on YouTube have exceeded 600,000 in less than two years. The volume has exceeded 300 million times. With the advancement of AI technology and the application of blockchain in payments, value transfer and open data platforms, these agents are expected to become more autonomous and may open a new mainstream entertainment category by 2025. -Katie , Dmitriy

Clockwise from top left: Hatsune Miku, Luna from Virtuals, Lil Miquela and PLAVE

4. Generative/Agentic Content Marketing

In some cases, the agent is the product itself, while in other cases, the agent can be a complement to the product. In the attention economy, consistently outputting engaging content is key to the success of any idea, product or company. Generative/agent-driven content gives teams a powerful tool to ensure a scalable, 24/7 content creation pipeline. This area has been accelerated by discussions on “memecoins vs. agents.” Agents are a powerful tool for memecoins to spread, even if they are not fully "intelligent" yet.

As another example, the gaming industry is increasingly pursuing dynamism in order to keep users engaged . A classic approach is to lead user-generated content, and purely generative content (such as in-game items, NPCs, or even fully generated levels) may be the next stage of this trend. We’re curious to see how the capabilities of agents in 2025 will further push the boundaries of content distribution and user interaction. -Katie

5. Next-Gen Art Tools/Platforms

In 2024, we launched the IN CONVERSATION WITH series, an interview program that talks to crypto artists in music, visual arts, design, curation and more. This year’s interviews made me notice a trend: artists interested in crypto are often also passionate about cutting-edge technologies and want these technologies to be more deeply integrated into their creative practices, such as AR/VR objects, code-generated art, and real-time coding (livecoding).

The combination of generative art and blockchain technology has been around for a long time, which also makes blockchain an ideal carrier for AI art. Displaying and presenting these art forms is very difficult in traditional platforms. ArtBlocks provides a preliminary exploration of how digital art can be displayed, stored, monetized and preserved through the blockchain, greatly improving the experience of artists and audiences. In addition, AI tools allow ordinary people to easily create their own works of art . We are very much looking forward to how blockchain will further enhance the capabilities of these tools in 2025. -Katie

KC : Since you are frustrated and disagree with crypto culture, what motivates you to still choose to participate in Web3? What value does Web3 bring to your creative practice? Is it experimental exploration, financial reward, or something else?

MM : For me, Web3 has had a positive impact on me personally and on other artists in many ways. Personally, platforms that support publishing generative art are particularly important to my practice. For example, you can upload a JavaScript file, and when someone mints or collects a piece, the code will run in real time and generate a unique work of art in the system you designed. This process of real-time generation is a core part of my creative practice. Introducing randomness into the systems I write and build has profoundly impacted the way I think about art, both conceptually and technically. However, it is often difficult to convey this process to an audience if it is not presented on a platform specifically designed for this art form, or in a traditional gallery.

In a gallery, an algorithm might be shown running in real time via a projection or screen, or works selected from multiple outputs generated by the algorithm and somehow translated into physical form for exhibition. But for audiences less familiar with code as an artistic medium, it can be difficult to grasp the significance of this randomness in the creative process, which is an important part of the practice of all artists who use software in a generative way. . I sometimes find it difficult to emphasize the core idea of ​​code as a creative medium to the audience when the final form of the work is just a picture posted on Instagram, or a physical print. .

The emergence of NFT excites me because it not only provides a platform to showcase generative art, but also helps popularize the concept of "code as an artistic medium" so that more people can understand the uniqueness and value of this way of creation.

Excerpted from IN CONVERSATION WITH: Maya Man

6.Data Markets

Ever since Clive Humby proposed the idea that “data is the new oil,” companies have taken steps to hoard and monetize user data. However, users are gradually realizing that their data is the cornerstone of the survival of these giant companies, but they have little control over how the data is used and fail to derive benefits from it. With the rapid development of powerful AI models, this contradiction has become increasingly acute. On the one hand, we need to solve the problem of user data being misused; on the other hand, as larger-scale and higher-quality models exhaust the "resource" of public Internet data, new data sources are also particularly important.

In order to return control of data to users, decentralized infrastructure provides a broad design space. This requires innovative solutions in multiple areas such as data storage, privacy protection, data quality assessment, value attribution and monetization mechanisms. At the same time, in response to the shortage of data supply, we need to think about how to use technological advantages to build competitive solutions, such as creating higher value data products through better incentive mechanisms and filtering methods. Especially in the current context where Web2 AI still dominates, how to combine smart contracts with traditional service agreements (SLA) is a direction worthy of in-depth exploration. - Danny

7. Decentralized Compute

In the development and deployment of AI, in addition to data, computing power is also a key element. Over the past few years, large data centers have dominated the development of deep learning and AI, relying on exclusive access to space, energy, and hardware. However, with the limitations of physical resources and the development of open source technology, this pattern is gradually being broken.

The computing v1 stage of decentralized AI is similar to Web2's GPU cloud, but it has no obvious advantages in terms of hardware supply and demand. In the v2 stage, we saw some teams starting to build a more complete technology stack, including high-performance computing orchestration, routing, and pricing systems, while developing proprietary features to attract demand and improve inference efficiency. Some teams focus on optimizing inference routing across hardware via compiler frameworks, while others develop distributed model training frameworks on their compute networks.

In addition, an emerging market called AI-Fi is forming, which uses innovative economic mechanisms to convert computing power and GPUs into income-generating assets, or uses on-chain liquidity to provide new ways of financing hardware for data centers. However, whether decentralized computing can truly realize its potential still depends on whether the gap between ideas and actual needs can be bridged. - Danny

8. Compute Accounting Standards

In decentralized high-performance computing (HPC) networks, how to coordinate heterogeneous computing resources is an important challenge, and the current lack of unified accounting standards makes this problem more complicated. The output results of AI models are diverse, such as model variants, quantization, randomness through temperature and sampling hyperparameter adjustment, etc. In addition, different GPU architectures and CUDA versions will also lead to differences in hardware output results. These factors make how to accurately statistically model and calculate market capacity in heterogeneous distributed systems an urgent problem to be solved.

Due to the lack of these standards, we have repeatedly seen in the Web2 and Web3 computing markets this year that model performance and the quality and quantity of computing resources have been miscalculated. This forces users to verify the actual performance of AI systems by running their own benchmarks or limiting the usage rate of the computing market.

The field of encryption has always emphasized "verifiability", so we hope that by 2025, the combination of encryption and AI will make system performance more transparent. Regular users should be able to easily compare key output characteristics of a model or computing cluster to audit and evaluate the actual performance of the system. - Aadharsh

9. Probabilistic Privacy Primitives

Vitalik mentioned a unique contradiction in the article "The Promise and Challenges of Crypto + AI Applications" : "In cryptography, open source is the only way to achieve security, but in AI, making the model (or even the training data) public will greatly greatly increases its risk of adversarial machine learning attacks.”

Although privacy protection is not a new research direction in blockchain, with the rapid development of AI, the application of privacy-related cryptography technologies is accelerating. This year has seen significant progress in privacy-enhancing technologies such as zero-knowledge proofs (ZK), fully homomorphic encryption (FHE), trusted execution environments (TEE), and multi-party computation (MPC). These techniques are used in scenarios such as private shared state for general purpose computation on encrypted data. At the same time, technology giants like Nvidia and Apple are also leveraging proprietary TEE technology to achieve federated learning and private AI inference while keeping hardware, firmware, and models consistent.

In the future, we will focus on how to preserve privacy in random state transitions, and how these techniques can facilitate practical applications of decentralized AI on heterogeneous systems, such as decentralized private inference, storage and access pipelines for encrypted data, and Construction of a fully autonomous execution environment. -Aadharsh

Apple 's Apple Intelligence stack and Nvidia's H100 GPU

10. Agent Intents and Next-Gen User Trading Interfaces

An important application of AI agents is to help users complete transactions autonomously on the chain. However, in the past 12-16 months, the definitions of terms such as "agent intention", "agent behavior", and "solver" have been vague, and the difference from traditional "robot" development is not clear enough.

In the coming year, we expect to see more sophisticated language systems combined with multiple data types and neural network architectures driving the field forward. Will agents continue to use existing on-chain systems to complete transactions, or will they develop entirely new tools and methods? Will large language models (LLMs) remain at the core of these systems, or will they be replaced by other technologies? At the user interface level, will users interact with the system through natural language to complete transactions? Will the classic “wallet as browser” theory become a reality? These are questions worth exploring. - Danny , Katie , Aadharsh , Dmitriy