In February of this year, the AI protocol Virtuals within the Base ecosystem announced its cross-chain integration with Solana. However, the crypto market subsequently entered a liquidity crunch, causing the AI Agent sector to shift from a bustling environment to a lull, and the Virtuals ecosystem fell into a period of dormancy.

At the beginning of March, BlockBeats conducted an exclusive interview with empty, the co-founder of Virtuals. At that time, the team had not yet launched the widely discussed Genesis Launch mechanism, but they were already exploring internally how to activate old assets, increase user engagement, and reconstruct token issuance and financing paths through mechanism design. It was a time when the market had not yet recovered, and the ecosystem was still in a cold start phase, yet the Virtuals team did not stop their efforts; instead, they were actively seeking new product directions and narrative breakthroughs.

Two months later, the AI Agent sector has warmed up again, with the Virtuals token rebounding over 150%. The Genesis mechanism has become an important trigger for revitalizing the ecosystem. From the dynamic adjustment of point acquisition rules to the continuous rise in project participation enthusiasm, and the mechanism loop of "new tokens bringing old tokens," Virtuals has gradually emerged from the winter and once again become a focal point of discussion.

It is worth noting that the Genesis mechanism of Virtuals shares some similarities with the Alpha points system recently launched by Binance, which evaluates user participation in the Alpha and Binance wallet ecosystems to determine eligibility for Alpha token airdrops. Users can earn points through holding and trading, with higher points increasing their chances of participating in new projects. By filtering users and allocating resources through the points system, project teams can more effectively incentivize community participation, enhancing the fairness and transparency of the project. The explorations of Virtuals and Binance may signal the emergence of a new trend in crypto financing.

Looking back at this conversation, the ideas and judgments expressed by empty during the interview are gradually revealing their foresight. This is not just an interview about the new mechanism; it is also a deep discussion about the path construction and underlying logic of "asset-driven AI protocols."

From "Product" to "Platform": Wall Street-style Infrastructure for AI Agents

BlockBeats: Can you briefly share what the team has been focusing on recently?
empty: Currently, our work is mainly focused on two parts. The first part is that we hope to build Virtuals into a service platform similar to "Wall Street" for agents. Imagine if you are an entrepreneur focused on building agents or agent teams; the entire process from financing, token issuance to liquidity exit requires systematic support. We want to provide a complete service system for teams that are truly focused on agent and AI research and development, allowing them to concentrate on developing their underlying capabilities without being distracted by other aspects. This work also includes content related to retail trading, which we can elaborate on later.

The second part is that we are deeply advancing our AI-related layout. Our vision is to build an AI society where every agent can focus on their strengths while achieving greater value through collaboration with each other. Therefore, we recently released a new standard—ACP (Agent Communication Protocol)—aimed at enabling different agents to interact and collaborate to jointly promote their business goals. These are the two main directions we are currently pursuing.

BlockBeats: Can you elaborate on that?

empty: In my view, our target customer groups can be divided into three categories: the first category is teams focused on developing agents; the second category is investors, including retail investors, funds, and various investment institutions; the third category is end-users, which are individual users who ultimately use agent products.

However, our main focus is actually on the first two categories—teams and investors. For the end-users, we do not intend to intervene directly but hope that each agent team can address the expansion of the end-user market themselves.

Additionally, we believe that interaction between agents should become a core model. Simply put, future services should be more about one agent selling or providing to another agent rather than purely selling to human users. Therefore, in our team's business development efforts, we are actively helping existing AI teams find such clients and collaboration opportunities.

BlockBeats: Can you provide some specific examples?

empty: "Wall Street" is essentially about building a capital operation system. Suppose you are a tech team looking to raise funds; the traditional path is to seek VC funding, and after securing the funds, you start to develop. If the project performs well, you might consider entering the secondary market, such as listing on the New York Stock Exchange or issuing tokens on exchanges like Binance for liquidity exit.

We hope to streamline this entire process—from early financing to the flexible use of funds during project development, and finally to liquidity exit in the secondary market. This is a complete chain we aim to fill.

This part of the work is different from ACP (Agent Communication Protocol), which is more about establishing interaction standards between agents and does not directly involve the capital operation system.

BlockBeats: How does it differ from the current Virtuals Launchpad? Is the funding also coming from the end-users?

empty: Actually, if you issue a token on Virtuals now without truly raising funds, it’s just issuing a token without real capital. The service we currently provide is to set a trading tax mechanism during buy-sell transactions, extracting a portion of the tax to return to entrepreneurs, hoping this can become a source of cash flow for them.

However, the issues can be divided into two parts. The first is how to truly help teams complete financing, which we have not yet fully resolved. The second is the structural issues inherent in the current project issuance model. Simply put, the current version resembles the old Pumpfun model—when a project is launched, some tokens are sold to external investors. But the reality is that there are too many institutional groups and "snipers" in the market.

When a truly excellent project issues a token, it is often snatched up by institutions at a high valuation before it can reach ordinary retail investors. By the time retail investors can access it, the price is often already high, and the project quality may have deteriorated, distorting the entire value issuance system.

To address this issue, we hope to explore a new token issuance and financing model aimed at ensuring that the project’s tokens are neither tightly held by the project team nor preferentially directed to large institutions in the English-speaking world, but rather genuinely left for ordinary investors who believe in the project and are willing to support it long-term. We are contemplating how to design such a new issuance mechanism to solve this fundamental problem.

BlockBeats: What would the specific ideas for the new model look like?

empty: Regarding funding, we have not fully figured it out yet. At this stage, the most direct way is still to seek VC funding or use public pre-sales for fundraising. However, to be honest, I personally do not particularly endorse the traditional public pre-sale model.

In terms of "fair issuance," we are trying to think from a different angle—hoping to redesign the mechanism based on "reputation."

Specifically, if you contribute to the entire Virtuals ecosystem, such as early participation, providing support, or building, you will have higher priority when purchasing quality tokens later. Through this approach, we hope to allocate resources more to users who genuinely support the ecosystem's development rather than being dominated by those looking for short-term arbitrage.

How to "self-sustain" the team from trading taxes

BlockBeats: Would you consider adopting a model similar to the LBP model previously launched by Fjord Foundry, or a whitelist mechanism like Daos.fun? These models are somewhat similar to your earlier idea of "those who contribute to the ecosystem enjoy priority." However, such practices have also sparked some controversies, such as insider manipulation and unfair distribution. Will Virtuals consider leveraging the advantages of these models in its design or specifically avoid similar issues?

empty: I believe the biggest problem with the whitelist mechanism is that the selection power of the whitelist is in the hands of the project team. This is very similar to "mouse warehouse" behavior. The project team can choose to allocate whitelist spots to their own people or friends, resulting in the final tokens still being held by a few individuals.

What we want to do is still a mechanism similar to a whitelist, but the difference is that the acquisition of the whitelist should be based on a publicly transparent rule system rather than being unilaterally decided by the project team. Only in this way can we truly achieve fair distribution and avoid insider manipulation issues.

I believe that in today's AI era, many times, starting a business does not require a large amount of capital. I often emphasize to the team that they should prioritize self-sufficiency, such as building a community, rather than thinking about financing from the outset. Because once you seek financing, it essentially means taking on debt.

We prefer to view the early development path from the perspective of Training Fees. That is to say, projects can choose to issue tokens directly and support daily operations through cash flow generated from trading taxes. This way, projects can obtain initial funding during the public construction process without relying on external investment. If the project grows, there will naturally be opportunities for liquidity exit in the secondary market.

Of course, the ideal situation is for the project itself to have a stable source of cash flow, so that they do not even need to sell their own tokens, which is the truly healthy and sustainable state.

I often share this idea with the team, and interestingly, those projects that are genuinely looking to "make quick money" lose interest upon hearing about this mechanism. They feel that in this model, there is no way to manipulate the mouse warehouse, nor is it easy to achieve short-term arbitrage, so they quickly choose to leave.

But from our perspective, this is actually a very good filtering mechanism. Through this approach, projects with different philosophies will naturally be filtered out, leaving behind those willing to genuinely build and align with our values to work together.

BlockBeats: This concept could develop some AI agents that can generate revenue.

empty: I think this is very necessary. To be frank, looking at today's market, there are very few products that truly have stable cash flow, but I believe that does not mean we should stop trying. In fact, among the teams we are connecting with every day, more than half still hold long-term visions. Many times, they have even provided us with VC-stage funding support in advance or expressed strong willingness to collaborate.

For them, the real goal is to cultivate a good community because the community can provide better feedback for their products. This may sound a bit absurd, but there are indeed many such teams, and those are the teams we genuinely want to support.

Who Should AI Agents Be Sold To?

BlockBeats: You just mentioned this "AI Wall Street" product system—from financing, issuance to exit, it constructs a complete process. Is this mechanism more aimed at incentivizing teams that are willing to issue tokens? Or does it also consider how to better support teams that hope to develop through the cash flow of their products? Will these two types of teams be treated differently in your system, or is there any mechanism design that can reasonably support entrepreneurs on different paths?

empty: Yes, the core responsibility of our business development is actually to encourage teams to issue tokens. To put it directly, it is to guide them to think about the possibilities and significance of issuing tokens. Therefore, the most common question teams ask is, "Why should we issue tokens?" At this point, we need to adopt different methods and perspectives to help them understand the underlying value logic. Of course, if the final judgment is that it is not suitable, we will not force them to proceed.

However, we have observed a very obvious trend: the traditional financing path has become increasingly difficult to navigate. The model of raising funds to grow and then issuing tokens has gradually become ineffective. Faced with this reality, many teams find themselves in an awkward position. We hope to provide a different solution from the on-chain and crypto perspective, allowing them to find new development paths.

BlockBeats: I understand. What I was trying to express earlier is that you also mentioned that traditional AI models largely still rely on "burning money" competition. However, after the emergence of DeepSeek, some smaller teams or investors in the market have begun to regain confidence and are eager to enter this field. What do you think of this phenomenon? Will it have a certain impact on teams currently engaged in AI foundational research or AI application layer development?

empty: Yes, I think if we set aside DeepSeek for now, from a traditional perspective, up to this point, the only company truly making money in the AI field is NVIDIA; almost all other players have not yet achieved profitability. So, in reality, no one has truly enjoyed the benefits of this business model, and everyone is still exploring how to create truly productive applications for the C-end market.

No other field can receive community feedback as quickly as the crypto space. Once you issue a token, users will actively read every word of the white paper and try every feature of your product.

Of course, this mechanism does not suit everyone. For example, some Agent products are more Web2-oriented, and crypto users may not perceive their value. Therefore, I also encourage teams working on Agents to seriously consider how to truly leverage and design Crypto as a differentiating factor in their products within the Virtuals ecosystem.

BlockBeats: I particularly agree with this point. In the crypto field, the iteration speed of AI is indeed very fast, but do the feedback provided by this group of users truly represent real market demand? Or do these feedbacks really align with more mainstream, scalable needs?

empty: I think many times, the product itself should not be forcibly promoted to unsuitable user groups. For example, one of the most successful aspects of AIXBT is that its users are precisely those who speculate on others' content, so their usage behavior is very natural and does not feel like they are being forced to use a boring product. The concept of mass adoption has been discussed for many years, and perhaps everyone should have given up on this obsession long ago. We might as well accept it and just sell to crypto users.

BlockBeats: What kind of dynamic relationship should exist between AI Agents and the tokens corresponding to them?

empty: Yes, I think this can be divided into two core points. First, it is not about investing in a specific AI Agent, but rather investing in the team that operates that Agent. You should understand it as a mindset closer to venture capital: you are investing in the person, not the product they are currently working on. Because the product itself can change rapidly; a month later, the team may realize that the direction is wrong and adjust immediately. Therefore, the "token" here essentially represents trust in the team, rather than a specific Agent itself.

The second point is the expectation that once a certain Agent product is developed, it can truly generate cash flow or have actual use cases (utility), thereby enabling the corresponding token to have empowering effects.

BlockBeats: What empowering methods do you think are currently not seen but may emerge in the future and are worth looking forward to?

empty: There are actually two main areas. The first is the more common model where you must pay to use my product or use tokens for payment, thereby indirectly achieving "soft destruction" or consumption of the tokens.

But I think a more interesting empowering method is to think from the perspective of customer acquisition costs. In other words, you want your users to also be your investors, so they have the motivation to actively help you promote and attract more users.

Open Source ≠ Empowerment, Developers ≠ Community

BlockBeats: Based on these views, what do you think of ai16z? In terms of project design and token mechanisms, their overall performance seems rather pessimistic?

empty: From a very pure investment perspective, setting aside our relationship with them, it is quite simple. What they are doing now does not empower the token itself at all. From an open-source perspective, an open-source model cannot directly empower a token.

However, the reason it still holds value is that it acts like a call option; that is to say, if one day they suddenly decide to do something, like launch a launchpad, then those who knew in advance and participated early may benefit from it.

Developers may indeed use their Launchpad in the future, and only at that moment will the token truly generate empowerment. This is currently the biggest question mark—if this model really works, I believe it will indeed be very powerful because they have reached a large number of developers.

But I still have many doubts. For example, even if I am a developer using Eliza, it does not mean I will necessarily choose to issue tokens on their Launchpad. I will compare options. Moreover, the product capabilities and community operation capabilities required to create a Launchpad are completely different from those needed to create an open-source framework, which is another important uncertainty.

BlockBeats: In what ways is this difference manifested?

empty: At Virtuals, we deal with customer service-related issues almost every day. As long as any team experiences a rug on our platform, even if it is not directly related to us, users will come to us to complain immediately.

At this point, we must step in to reassure users and think about how to reduce the overall risk of rugs. If a team is hacked or has assets stolen due to incorrect token design or technical errors, we often need to dig into our own pockets to ensure that their community can at least recover some funds so that the project can restart. These project teams may be technically strong but may not be adept at token issuance, resulting in asset loss due to operational errors. As long as it involves issues related to "being deceived," it becomes very troublesome for us; doing this work is not much different from customer service at an exchange.

On the other hand, business development is also very challenging. Excellent teams have many choices; they can choose to issue tokens on Pumpfun or exchanges. Why would they come to us? There must be a complete support system behind this, including financing support, technical assistance, market promotion, etc. Every link must not have issues.

BlockBeats: Let's continue along this topic and discuss the current Launchpad business of Virtuals. Some community members have compiled the overall profitability status of the Virtuals Launchpad on Twitter, and it does seem that the number of profitable projects is relatively low at the moment. Will Launchpad still be a major business segment for Virtuals going forward, or will the focus gradually shift to the "AI Wall Street" path you mentioned earlier?

empty: In fact, these two aspects are essentially one thing; they are part of a complete system, so we must continue to push forward. Market fluctuations are very normal. One point we must always adhere to is to clearly recognize who our core customers are. I have always emphasized that we only have two types of customers—teams. Therefore, the state of the market is not the most important thing for us; the key is whether, at every critical juncture, the best choice for a team to issue tokens is still us, Virtuals.

BlockBeats: Are you worried that the narrative of "Crypto + AI" or "Crypto AI Agent" has already passed? If there is another bull market in the future, do you think the focus of market speculation may no longer be on these directions?

empty: It is possible. I think it is what it is; this could indeed happen, but it is also beyond our control. However, if you ask me which track among all possible trends has a better chance of maintaining a long-term lead, I still believe it is AI. From the perspective of playing poker, it remains the optimal choice.

Moreover, our team's technical framework and underlying capabilities have already been established; we are just going with the flow now. More importantly, we genuinely love this work and approach it with curiosity. Waking up every morning with the drive to study the latest technologies is itself quite satisfying, right?

Many times, people should not only look at the product itself. In fact, many excellent teams have genes that determine their ability to win within the rules—they may have previously engaged in proprietary trading with each transaction being in the millions, and the CEOs of these teams might earn a salary of 1 million dollars a year. If they are willing to go out and start their own projects, from the perspective of angel investment or VC, this essentially means acquiring a high-quality team at a very reasonable price.

Moreover, these assets are liquid, not locked. If you are not in urgent need of money, you can completely buy some tokens from excellent teams in the early stages and quietly wait for them to create some miracles. That’s basically the logic.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。