ZetaChain, together with its multi-model aggregation application Anuma, creates a new AI experience by building a "privacy memory layer" across large AI models.

Author: Zen, PANews

In today's generative AI applications, users often face fragmented conversational experiences. When switching between different models, the context of previous conversations is often lost, requiring information to be repeated from the beginning each time. For example, project details discussed on ChatGPT cannot be directly inherited when switching to Claude or other models, severely impacting efficiency.

Moreover, the dialogue data from these large models is typically stored on the servers of various platforms, leaving users with neither privacy protection nor control over their own data. "These real-world problems not only create a fragmented user experience but also raise concerns about user data sovereignty and security."

To address this pain point, the industry has begun exploring the concept of a "portable, user-controlled memory layer," and blockchain technology may be the key to achieving this goal.

Based on the open interoperability of blockchain, it may be possible to create a privacy memory layer to store AI context as a digital asset and transfer it seamlessly between multiple AI platforms, so that we no longer have to worry about "forgetting" past communication content every time we change tools, while ensuring data privacy and sovereignty.

ZetaChain 2.0 Released, Creating a Universal Layer for AI and Web3

In response to these needs, ZetaChain, a public blockchain project focused on cross-chain interoperability, seized the opportunity presented by the convergence of AI and Web3. In its roadmap review at the end of 2025, ZetaChain announced its "2.0" version plan, which will introduce new features for the AI era on top of its existing general cross-chain architecture.

On January 27, 2026, ZetaChain 2.0 was officially launched, along with its first AI product—Anuma, a large-scale model aggregation application with privacy at its core. According to the official introduction, ZetaChain 2.0 revolves around the following three core capabilities:

The Private Memory Layer is a protocol-level memory system specifically designed for AI interaction. It aims to bridge the contextual gap between AI tools, allowing users to truly control their digital memories. Based on the Private Memory Layer, all user conversations are encrypted and stored, with the key held only by the user; the platform itself has no access. Valuable information generated at different times and in different models is also under the user's control, continuously accumulating and readily migrated to new conversations, rather than being monopolized by any single AI service.

AI Portal is a unified routing and execution layer that enables applications to access multiple AI model providers without being locked into a single platform. It also features built-in support for availability, fallback, and cost/performance optimization. AI Portal handles the underlying model routing and contextualization. Users can freely choose different models such as ChatGPT, Anthropic Claude, and Google Gemini to obtain answers, with previous conversations supported by a privacy memory layer.

In addition to the protocol itself, ZetaChain 2.0 also presents key capabilities as a developer toolkit (SDK). Developers can directly integrate privacy-preserving memory, cross-model switching, and monetization components into their products. This SDK allows applications or AI agents to maintain continuous context across different models and invoke different model capabilities on demand, thereby significantly reducing the cost and complexity of teams building their own infrastructure.

From a mechanism design perspective, the three core modules mentioned above complement each other. The private memory layer provides privacy-preserving user memory and data support, the AI portal offers the ability to continuously interact with mainstream large models, and the SDK ecosystem facilitates efficient and quick participation and expansion by third-party developers. This allows ZetaChain to expand from an underlying cross-chain protocol into a general-purpose platform that simultaneously serves Web3 and AI.

With privacy and user sovereignty at its core, Anuma has been launched and is now open for applications.

Alongside the official launch of ZetaChain 2.0, another major highlight from the project team was Anuma, the platform's first consumer-grade AI product. Currently, Anuma is in its private beta phase, with access gradually being granted through an invitation-only waiting list. Users can apply for early access by participating in the public waiting list.

As a large model aggregation application, Anuma integrates with multiple mainstream large models, allowing users to invoke different AI engines within a single conversation. It offers the convenience of aggregation tools like PoE, while also supporting OpenAI's GPT series and Anthropic's Claude.

When a user asks a question, they can specify or change the model used for the answer. Switching between engines is as simple as clicking a selection button, without needing to migrate to another application. Users can flexibly choose the most suitable model to obtain answers on Anuma based on the question type, while the entire dialogue process continues continuously within the same window.

On the technical level, thanks to ZetaChain's private memory layer, every conversation a user engages in within Anuma is encrypted and stored as personal memory, seamlessly migrating to new models or conversations. When a user initiates a new conversation or switches AI models within an existing one, Anuma can securely inject the relevant context into the target model, enabling it to understand the previous background and user intent. This eliminates the need for users to repeatedly explain the same background information across different AIs, significantly improving the efficiency of cross-model collaboration.

Traditional Web2 companies have long been abhorrent for exploiting user data for malicious purposes, leveraging their centralized advantages. Practices such as price discrimination against loyal customers and the sale of data persist despite repeated crackdowns. This wariness and anxiety among users regarding centralized platforms has extended to the rapidly developing field of AI.

Anuma places great emphasis on the privacy of conversations and user control, employing an end-to-end encryption scheme to protect user data throughout the platform. From the moment a user enters a message on the front end, the content is encrypted using the user's key before being passed to a privacy memory layer for storage. When the context needs to be provided to the AI model, it is decrypted by the user or a trusted execution environment before being sent to the model. Throughout the entire process, the conversation record is always stored in encrypted form on the blockchain or during transmission, and even ZetaChain nodes or servers cannot view the content.

This contrasts sharply with traditional AI chat services, whose chat logs are typically stored in plaintext on servers, raising concerns about potential access or leakage by operators. Anuma, however, utilizes blockchain and encryption technology to achieve a level of security similar to Web3 wallet private key management, ensuring that user data can only be accessed by the user. This provides a more secure option for AI applications in sensitive scenarios such as law and healthcare, encouraging users to engage in more private conversations.

In fact, before Anuma was launched, there were already some AI dialogue products that aggregated multiple models on the market. Typical examples include the Poe platform launched by Quora, the "American version of Zhihu", and TypingMind from the open source community.

Compared to the cloud service models and localized deployments of these two platforms, Anuma's on-chain encrypted storage balances privacy and sovereignty. In terms of ease of use and model richness, Anuma eliminates the cumbersome configuration process of the TypingMind model, providing a convenient multi-model dialogue experience similar to PoE.

Behind ZetaChain's foray into AI: Its technological logic and natural evolution

The ZetaChain team's decision to launch version 2.0 and Anuma at this time is actually backed by deep technical expertise and a clear evolutionary logic.

As the first universal L1 blockchain project, ZetaChain has focused on overcoming the fragmentation problem in the blockchain field since its launch in 2021, aiming to build an underlying network connecting all public chains. Built on the Cosmos SDK, it naturally supports interoperability with heterogeneous chains such as Ethereum, Bitcoin, and Cosmos.

ZetaChain simplifies traditional cross-chain operations, which require bridging and encapsulation, into a single contract call on a single chain through innovations such as CAF, providing users with unified liquidity and a seamless user experience. As of the end of 2025, the ZetaChain mainnet had integrated with ten major blockchain networks, including Bitcoin, covering tens of millions of users, with a cumulative on-chain transaction count of 225 million.

ZetaChain has also gained widespread recognition in terms of its ecosystem and capital. According to public data, the project has secured $27 million in funding from well-known institutions such as Blockchain.com, Jane Street, and Sky9 Capital. In 2024-2025, global technology and infrastructure giants such as Google Cloud, Deutsche Telekom, and Alibaba Cloud joined the network as validator nodes, endorsing its security and compliance.

Entering the second half of 2025, with the explosion of generative AI, the ZetaChain team realized that there were similar pain points between the industry's multi-chain ecosystem and multi-model AI: fragmentation across multiple platforms and systems, requiring a universal layer for integration. Therefore, they proposed the strategic concept of an "AI Universal Platform," introducing the trusted computing and storage capabilities of blockchain into the AI field to create blockchain infrastructure for the AI era.

ZetaChain 2.0 is the realization of this vision. It retains and enhances the original cross-chain functionality while adding AI privacy memory and interaction capabilities. This aligns with ZetaChain's consistent vision of making Web3 equally friendly to both humans and AI. The natural evolution from a "general-purpose blockchain" to an "AI general-purpose platform" is both in line with the general trend of technological convergence and an extension of the project's mission.

"ZetaChain has achieved scalable unification at the blockchain experience level." As Ankur Nandwani, a core contributor to ZetaChain, stated, ZetaChain 2.0 extends the same approach to AI, enabling next-generation applications and agents to run between models and the blockchain, and possessing by default private, authoritative memory capabilities and global monetization channels.

What are the prospects for this new paradigm of deep integration between blockchain and AI?

The launch of ZetaChain 2.0 and its debut product, Anuma, represent a significant attempt to deeply integrate blockchain and AI. Within this system, we witness a new paradigm for multi-model AI applications: privacy-first, user-controlled, and cross-platform compatibility.

Of course, it should be objectively pointed out that Anuma is still in a very early Private Beta stage, and its overall ecosystem is still under construction. Many features and details still need feedback from testers to improve, such as support for more models, memory layer capacity and performance optimization, and the enrichment of third-party developer tools. This means that Anuma is far from being able to replace the experience of a mature single platform in the short term, and some users will need time to get used to this new interaction mode.

However, it cannot be ignored that Anuma represents a groundbreaking approach. In the multi-model aggregation experience arena, Anuma offers a different approach from the solutions provided by major companies. Instead of a centralized platform monopolizing data and model access rights, it returns the power of choice and memory to the user, achieving coordination with minimal trust through blockchain technology.

As Anuma's public beta opens and its features iterate, more innovative applications may emerge on this platform, such as AI advisors to ensure privacy and intelligent search assistants across different fields. How far this new trend of privacy-first, multi-model experiences can go remains to be seen.