The market has long cycled between “AI is no longer hot” and “AI is hot again.” Especially with top VC investments, AI projects often go from lukewarm to widely discussed overnight. Yesterday, aside from the buzz around Vitalik’s live performance at Token 2049, the most talked-about project in the CT space was Vana. Following the project’s announcement of securing a total of $25 million in funding from the three crypto VC giants—Coinbase Ventures, Paradigm, and Polychain—posts with likes, congratulations, and analyses quickly flooded in.

You may not love VC coins, but in a market lacking in hot topics, any major funding round is still worth paying attention to. Vana’s Twitter bio reads: “User-owned AI through user-owned data.” It seems like yet another narrative against big company monopolies on AI, another focus on data, and once again, the usual VC giants gathering together. So, what sets Vana apart this time?

Data DAO sounds sophisticated and appealing, bringing together the VCs

With Vana’s funding announcement, optimism spread quickly. For example, one Twitter user pointed out that Vana is Paradigm’s first AI investment, and that Vana received backing from different top-tier VCs in various funding rounds:

• Strategic round: $5 million from Coinbase Ventures
• Series A: $18 million led by Paradigm
• Seed round: $2 million from Polychain

When the three VC giants jointly back a project, it’s usually because the project’s narrative and the problems it aims to solve are significant. As mentioned earlier, Vana’s focus is on the “dragon-slaying” narrative of data control in the AGI era—individuals contribute data without reward, and data privacy is not guaranteed; meanwhile, large companies dominate the training of AI models. Thus, the goal is to break this status quo.

However, following the AI trends of the past few years, advocating for multiple AI models and opposing monopolies has already become a narrative widely accepted in the crypto space.

(For related reading: Delphi Labs: AI Will See Multi-Model Competition, Which Crypto Applications Do We Favor?)

This isn’t a new story, but most current crypto AI projects telling this story are focused on the DePIN sector, encouraging the use of different hardware to contribute various resources.

Vana’s approach to solving the problem feels like an old idea with a new twist—using DAOs with different purposes to allow people to contribute different types of data, which can then be used to train AI models for different purposes.

In Vana’s technical blog post, they wrote:

“A Data DAO is a decentralized entity that allows users to pool and manage their data… It’s somewhat like a data union. The DAO has full control over the dataset and can choose to rent or sell anonymous copies. For example, Reddit’s data could even be used on new user-owned platforms, including friends, past posts, and other data, ready for use on these new platforms at any time.”

Currently, Vana’s official website lists 16 different data DAOs, allowing users to contribute various types of data such as from Reddit, Twitter, and dating apps, while giving them control over these data through Vana’s blockchain network.

At the same time, the data can be contributed to AI models that require specific vertical data for training, and users can earn rewards from this. If data from all sectors could be contributed through such DAOs, it would be an ideal model. However, the challenge lies in how to implement this.

At least two specific issues arise here:

1. How do you ensure that I am genuinely contributing data?
2. How do you guarantee the security and ownership of the contributed data?

This brings us to how Vana’s network operates and its underlying principles.

The core of Vana’s network lies in its unique multi-layer architecture, designed to create a decentralized data ecosystem that addresses key issues such as data ownership, privacy protection, and value creation.

According to the latest architecture diagram, the Vana network consists of three key components: the Data Portability Layer, the Data Liquidity Layer, and the Universal Connectome.

1. Data Portability Layer — This layer is the top layer of the Vana network, responsible for handling applications and AI models that utilize on-chain data. In this layer, there are mainly two types of participants:

• Developers: They create applications and AI models that use on-chain data.
• Users: They use these applications and models and are also the ultimate beneficiaries of the data.

The main function of this layer is to ensure data portability and interoperability, allowing users to easily transfer and use their data across different applications and models.

1. Data Liquidity Layer — This layer is responsible for bringing the data owned by users onto the chain and serves as the foundation of the entire system. It includes two important types of participants:

• Contributors: They contribute the data they own to the system.
• Validators: They are responsible for verifying the authenticity and value of the contributed data.

This layer addresses the issue of how to safely and reliably transform offline data into on-chain assets, providing liquidity for the entire ecosystem.

1. Universal Connectome — This is a unique component that acts as a real-time data flow map for the entire ecosystem. It consists of the following roles:

• Propagators: They are responsible for disseminating and updating information about data flow within the ecosystem.

The role of the Universal Connectome is to provide a real-time map of data flow throughout the ecosystem, enabling all participants to understand the direction and usage of data.

These three components work closely together to form a complete data ecosystem:

• The Data Liquidity Layer safely brings users’ data onto the chain, providing raw data for the entire system.
• The Data Portability Layer leverages this on-chain data to support the development and usage of various applications and AI models.
• The Universal Connectome monitors and displays the flow of data within the ecosystem in real time, ensuring transparency and traceability.

Without delving into the technical details of implementation, such as smart contracts, ZK, and code-level aspects, you can think of Vana as a blockchain network that offers a comprehensive solution for the contribution, verification, usage, and monitoring of data.

With elements like large infrastructure, the narrative of breaking the monopoly of AI models, and the innovative approach of DAOs, it makes sense that several top-tier VCs have come together to support this project.

(Note: Readers interested in the technology can directly visit the project documentation for more information.)

Click mining is grounded and unites the masses.

According to the previous tone and approach of crypto AI projects, such initiatives often focus on a high-end, tech-savvy narrative, leading typical users to feel impressed but unable to engage meaningfully. However, Vana has made its gameplay more accessible this time—alongside the highbrow narrative aimed at VCs, it also features a low-barrier, popular “click mining” model for retail investors.

Narrative is important, tone matters… but popularity is crucial.

Speaking of projects funded by Paradigm, Vana reflects the practices inherited from Blur and Friend.Tech, which involve user engagement through points and community-building. The main gameplay is straightforward:

1. Open Telegram and enter the Vana Data Hero mini-program (interested players can access it here).
2. The primary gameplay is quite simple; it involves clicking on icons bundled with various Web2 applications. The more you click, the faster you hit the screen, and the more VANA points you accumulate over a given time (symbolizing data contributions from various Web2 applications to mine points).

1. In addition to simply clicking the screen, there are various other activities like following social media, inviting friends, and adding the testnet, which can be completed to earn different scores.

The project team clearly understands marketing and how to ride trends. For instance, one task involves replying to a post by Elon Musk related to AI, data, and privacy, and completing it earns 500 VANA points. This feels like a way of leveraging the community and promoting Vana through association.

However, we don’t yet fully understand the conversion rate of these VANA points to tokens or the distribution rules. Interested players can follow Vana’s social media for more information. With the backing of major VCs and the success of similar Telegram mini-programs, it’s almost certain that token incentives and exchange listings are on the horizon. Still, the intense “PUA” points system can be both exhausting and FOMO-inducing, so how one chooses to participate will ultimately depend on individual preference.

Sir, times have changed.

The interesting part is that even a seemingly high-profile AI project is now using Telegram mini-programs to build hype and gain traction. In the past, AI projects often came across as high-tech and exclusive, but after the industry sparked discussions around VC coins not being picked up and the TON ecosystem mini-programs representing mass adoption, project teams are clearly adapting to the evolving landscape. They not only aim to align with VCs but also seek grassroots support.

If you look into Vana’s earlier background, you’ll notice that after its establishment in 2021, the project was initially positioned as an AI identity generation application.

It’s evident that building an application doesn’t carry the same weight as developing infrastructure, and focusing solely on a VC narrative isn’t as effective as appealing to both VCs and the general public. After Vana’s pivot, you can clearly see a shift in its Go-To-Market strategy. Grand narratives are limited, and new technologies don’t emerge instantly. Instead of waiting for market conditions to change, Vana has maximized its own adaptability to align with the current market rhythm.

I anticipate that more new projects, or revamped older projects, will continue to switch up their approaches. Timing and fate, as they say.

statement:

1. This article is reproduced from [TechFlow], the original title is “Can AI Survive in an Encrypted World: Encryption Experiments on 18 Large Models”, the copyright belongs to the original author [TechFlow], if you have any objection to the reprint, please contact Gate Learn Team, the team will handle it as soon as possible according to relevant procedures.

2. Disclaimer: The views and opinions expressed in this article represent only the author’s personal views and do not constitute any investment advice.

3. Other language versions of the article are translated by the Gate Learn team, not mentioned in Gate.io, the translated article may not be reproduced, distributed or plagiarized.