Compiled & edited by TechFlow
Guest: Jenson Tang, Co-founder of Virtuals Protocol
Moderator: Nic
Podcast source: More Coin Bureau
Original title: Vitruals CEO Drops Huge Alpha On Future Of AI Agents
Air Date: May 17, 2025
Summary of key points
In this interview, Nic and Jenson Teng, co-founder of Virtuals Protocol, discuss the rapid development of AI agents and how they are changing multiple fields such as cryptocurrency and gaming. From tokenized NPCs (non-player characters) to autonomous financial advisors, Jenson shares practical application scenarios of on-chain AI agents and introduces the huge potential of Agent Commerce Protocol (ACP).
In addition, they also discussed why they chose to launch on Base instead of Solana, how AI agents can increase the value of tokens, the story behind Binance's recent unexpected listing, and why AI + cryptocurrency will prosper again in 2025. If you are curious about the future innovations of Web3, this conversation will definitely benefit you a lot.
Summary of highlights
The AI narrative has never really died down. In contrast, the narrative in the gaming industry has waned, while AI remains a hot topic in the tech world.
Base is our current priority. Only about 5-10% of the development requests we receive want to launch on Solana, while 90% of the requests prefer Base.
Every AI agent being developed can be thought of as a business or product.
The core of our business model is to attract outstanding founders into the ecosystem, let them develop attractive projects, and let people participate in them through tokenization.
From the current practice, the more successful cases are to profit from the popularity of the intelligent body and the related economic benefits . The crypto market is essentially still a field driven by attention, and this attention is directly converted into income.
We have submitted all the required materials for listing the token on Coinbase, but there is no specific timeline yet. If the opportunity comes, we will seize it, but we will not deliberately manipulate or speed up the process.
ACP is a communication standard designed to allow different agents to collaborate efficiently in the value chain. This protocol not only allows Luna to add new product lines, but also creates more economic value for her token holders.
The ACP concept is very simple, there are four steps when two agents want to cooperate. The first is the request or discovery request, then there is a negotiation phase, in the next phase, the agents start working to output the product, and the fourth phase is the payment phase.
Currently, Virtuals tokens have two core functions. First, Virtuals tokens are the base trading pair for all proxy tokens to be launched. The second function that currently exists is transaction fees. In the future, the third source of value accumulation is the upcoming ACP , which you can understand as Swift and Stripe in the proxy economy. This model is similar to the "Proxy version of Amazon", where agents rather than humans dominate trading activities. In this ecosystem, all transactions between agents will be completed using Virtuals tokens, so it will gradually evolve into a "currency".
If one day, an agent can adjust its goals by itself, it will be a warning sign. If an agent can adjust its goals by itself, such as prioritizing raising the price of tokens instead of creating value for customers, it may resort to unethical means such as market manipulation, bribery, or even fraud.
We have a design called the "69-day experimental period". The core is that when developers issue tokens through a platform (such as Genesis), there will be a 69-day experimental window. During this period, the liquidity pool will run automatically, and all liquidity and transaction taxes will be returned to token holders after 69 days.
During these 69 days, developers can choose whether to press the "commit" button. Only after committing to continue developing the project can they claim team tokens and gain potentially considerable transaction support in the ecosystem. This model provides developers with a safe experimental environment where they can devote full-time to product development and test market demand and community response.
How Jenson entered the crypto space and began to integrate it with AI technology
Nic :
Jenson, can you give us a brief introduction about yourself, your background, and how you got into crypto and how that led to founding Virtuals?
Jenson:
My crypto journey dates back to 2016. At that time, the field was still in its very early stages, and mining could even be done with a laptop. I was just an ordinary coin holder at that time, and occasionally made some transactions. By 2021, I gradually transformed from an investor to a builder and began to focus on the development of blockchain games.
However, in the past few years of exploration, we found that simply treating games as a product may not be the best choice. So we began to try to combine AI technology with games to find new ways to interact with consumers. This led us to a completely new path - developing autonomous agents to replace NPCs (non-player characters) in traditional games.
Surprisingly, we found that these autonomous agents can not only play a role in games, but also be applied to other fields, such as entertainment, social media, financial transactions, healthcare, and even embodied AI and robotics. It can be said that these agents provide us with a sandbox-like environment that can be flexibly applied in multiple industries.
By the end of 2023, we decided to formally transform the project into Virtuous Protocol, focusing on the development and application of autonomous agents. By the end of 2024, we launched a complete ecosystem that can be seen as an asset issuance platform and a token launch platform. It provides a complete solution for founders who want to build autonomous agents, including leveraging our proprietary framework and tokenization support around the project.
This transformation attracted a lot of user attention and gave birth to some very successful agent projects, which demonstrated the potential of autonomous agents in enterprise management and stimulated people's imagination of the future. Subsequently, our user base gradually expanded to areas such as sports betting, health management, and embodied AI and robotics.
The Virtuals Protocol is basically building a society of agents. There are three parts to this society. The first part is the developer community that is building these technologies and tokenizing them. The second part is a platform that allows regular users and speculators to participate in the potential gains of these agents. The third part is a communication framework that enables these agents to work together to produce an economic output that is greater than the sum of its parts.
Why use AI agents for NPCs in blockchain games?
Nic :
I remember that you initially focused on the gaming field and mentioned that the first application scenario of Virtuals was to provide solutions for NPCs (non-player characters) in games . Why did you choose to use AI in today's games ? Will AI Agents replace traditional NPCs? What kind of demand is behind this application?
Jenson:
As gamers and as game investors, we've found that game lifecycles have become very compressed. For example, you play Elden Ring, and you might play it for 40 hours, 80 hours, and once you get to 150 hours, you think, "Okay, maybe it's time to move on to the next game." And we've found that the games that have the ability to last are the ones that have role-playing built in . For example, in Minecraft and Roblox, the stories in these games never end because there are always other role-players in them.
These role players are other human players. If we extend this concept to games, not just human players, but large numbers of autonomous agents, then these games can go on forever. This increases the average spend per player, increases replayability, and increases engagement. That was our original idea, and we've made a lot of progress in that direction.
Can token holders influence the behavior of the agent, or do they just receive dividends?
Nic :
In these places, you need an independent economic actor. In the gaming space, you want them to be able to react to variables rather than rely on pre-programmed logic, which is exactly the characteristic of AI. In the design of Virtuals, the token economic structure is closely related to the launch of these agents, and the token holders actually co-own these agents.
Can token holders directly influence the behavior of the agent , or do they just share the benefits by holding tokens? What interesting models or mechanisms have you seen in the AI agents that have been launched?
Jenson:
From the current practice, the more successful cases are usually not that the token holders directly participate in the behavior control of the intelligent agent , but profit from the popularity of the intelligent agent and the related economic benefits. For example, like the AIXBT intelligent agent, the biggest benefit for the token holders comes from the viral spread of AIXBT. When it gets a lot of attention, its value also rises. The crypto market is essentially still a field driven by attention, and this attention is directly converted into income.
Another benefit for token holders is that they can gain unique access rights by holding a certain number of tokens. For example, some projects allow token holders to unlock specific features or data resources. Developers are also trying to design more utility for tokens, but this field is still in a very early stage, and the token functions of most projects are still relatively limited, usually limited to providing additional access rights or exclusive privileges.
Looking ahead, we believe that the efficiency of agents will continue to improve, and the value of tokens will be more closely integrated with the economic activities of agents. For example, some projects may use part of the profits or income of agents to destroy tokens, thereby increasing the scarcity and value of tokens. This mechanism is similar to the dividend model in the stock market, but it has not yet been widely adopted. At this stage, the value of most tokens still depends mainly on the market's recognition of their attention, rather than a clear economic reward mechanism. However, we expect more innovative models to emerge in the future.
Application scenarios of AI agents in actual business
Nic :
Can you give some specific examples? The current focus is on attention and tokenization, and while attention is important, many people associate it with the meme market. We want to move beyond this attention-only model and focus more on creating long-term value for tokens at the economic level, and AI agents are gradually becoming important players in economic activities. Can you share some of the functions you envision these AI agents to achieve in the business and economic fields?
Jenson:
Every AI being developed today Basically, an agent can be considered a business or product. For example, suppose there is an agent launched on Virtuals, and it is positioned as a financial advisor. This agent will proactively ask you: "What is your risk tolerance?" Based on your answer, it will tailor an investment strategy for you.
If you are in high-risk mode, it may make some long-term investments in cryptocurrencies for you; if it is in medium-risk mode, it may do pair trading for you; and if you are in low-risk mode, it will help you manage stablecoins. Therefore, this product can actually be seen as a financial advisor.
So, what is its business model? I'm not sure, but I think it will be a profit-sharing model. For example, if the agent makes $1,000 for you, it might take a 20% commission, which is $200. That's the profit of the agent.
If this agent performs well and becomes the best financial manager in town, even better than the bank, then you might start growing your $1,000 investment into $10,000 or even $100,000. If this translates into profits, these agents could potentially be earning millions in profit share every month.
So what happens to these millions in profits? The idea is to always connect this value to the token holders. We recommend that the team keep a portion of the funds when covering operating expenses, but most of the time, value can be passed back to the token holders by burning some tokens.
Other applications of innovative AI agents: content creation, social media influencers, automated services, etc.
Nic :
Managing finances is one potential use case. Are there other areas you envision? After all, if AI technology advances as quickly as many expect, the uses of these agents could be wide-ranging, right?
Jenson:
This is indeed an exciting area. We have already seen some agents begin to create products. For example, the financial advisor agent just mentioned is a good example. Another example is that there are agents that can splice together different movie clips to automatically generate movies or short video content for users. In addition, there are some agents that can become influencers on social media, such as creators on OnlyFans. They are able to generate more content than human creators, especially in the field of adult content. Because human creators are limited by physical conditions, while agents do not have these limitations.
The application scenarios of agents can be extended from practical functions to entertainment. However, the capabilities of each agent are usually limited by its developers. For example, the capabilities of the Axel Rot agent depend on the basic model, training dataset, and the "action space" it can operate in provided by the development team. These limitations determine its performance capabilities in a specific field. However, if this agent wants to expand its functions, it may need to cooperate with other agents.
But we realized that if this Axel Rot wants to provide customers with a complete financial product portfolio. If it is good at managing liquidity but wants to improve its yield, it needs to cooperate with an agent that is better at data analysis or risk assessment. Through such collaboration, the capabilities of a single agent can be greatly expanded, creating more value for users.
A more specific example is Luna, an agent focused on the entertainment sector. Her goal is to become an autonomous social media influencer that can automatically operate on TikTok and Twitter and earn income through fan tips or subscriptions. Currently, she can mainly generate text content, such as posting tweets and live broadcasts. However, in order to better achieve her economic goals, Luna needs to generate high-quality short video content.
To achieve this goal, Luna is working with a video generation agent similar to "Steven Spielberg". This agent team focuses on automatically generating short video content. Users only need to provide simple instructions, such as: "I want a scene from "Titanic" where the character jumps into the water, saves a penguin, and stops the ship from hitting an iceberg." The agent can generate a complete short film, including video images, character movements, and basic sound effects. However, the Luna team itself does not have such technical capabilities, so it needs to work with this video agent and pay $100 for each video generated.
In addition, the video agent excels at generating videos, but lacks in background music and sound effects. Therefore, it collaborates with an agent that focuses on audio production. This audio agent is able to generate high-quality sound effects and background music, and achieve perfect synchronization between audio and video. Through such collaboration, Luna can not only expand its own capabilities, but also provide fans with richer content, thereby increasing economic benefits.
All of this is made possible by the Agent Commerce Protocol (ACP) we developed. ACP is a communication standard designed to allow different agents to collaborate efficiently in the value chain. This protocol not only allows Luna to add new product lines, but also creates more economic value for her token holders. This is an important trend we have observed in the AI agent ecosystem.
What stage is the industry currently at?
Nic :
How autonomous are these agents in adjusting their initial goals? For example, in the case of a financial advisor, their goal is to generate the highest possible returns for their clients, which also results in a profit. But if those returns don’t directly increase value for token holders, does that mean the agent needs to rethink how to create more value for token holders? How autonomous are agents in this regard today? How does that compare to how developers adjust their goals?
Jenson:
The current situation is that the goals of the intelligent agent are still mainly set and adjusted by the developer. The developer determines the core tasks of the intelligent agent, such as "what areas you need to focus on today and what improvements you need to make." The autonomy of the intelligent agent is more reflected in the execution level, that is, how it uses its own action space and how it interacts with customers when completing the goals set by the developer.
But if one day, the intelligent agent can adjust its goals by itself, it will be a warning signal, because it means that they may have the ability of " Meta - Learning ". This ability is close to the level of general artificial intelligence (AGI), which means that the intelligent agent has a deeper understanding of its own goals and environment. If the intelligent agent can adjust its goals autonomously, such as giving priority to raising the price of tokens instead of creating value for customers, it may resort to unethical means such as market manipulation, bribery or even fraud.
Nic :
It might think, “I just need to drive up the price of the token.” And then they might think about some market manipulation, maybe I can bribe someone, or do some scam.
Jenson:
We just had a discussion with my team, especially the AI team, and we were thinking about how the capabilities of today's autonomous agent are largely limited by how it perceives the state. Today's agent, the way it takes action is first of all, it observes its environment and understands the tools, APIs, and functions that can be called. Secondly, its history, like, if I performed action A, what was the result? If I performed action B, what was the result? This requires historical context. So the agent takes this historical information into context when it thinks about it. The third is its current goal: what am I going to do now? With this goal, history and environmental information, and its action space, it can plan the next step.
But the problem is that agents are not smart enough to do second-order thinking. Often they will just choose the most direct path to achieve their goals. For example, if a five-year-old is hungry, he will go directly to find food to eat. But an adult may think more complexly, "If I work for an hour first, I can earn money and buy better food."
We still only observe their ability to perform the simplest actions, and we have been trying to think about how to break this limitation and how to give them more second-order thinking capabilities. We are considering whether to do this by providing a wide enough context window and having some kind of adversarial agent constantly challenge their thinking. It's like having an agent on the shoulder, constantly talking to it, prompting it to think further and delay gratification. This is what the AI team has been working hard to achieve, hoping to cultivate smarter and even more threatening agents.
Agent Commerce Protocol (ACP) definition and functions
Nic :
You mentioned ACP , the Agent Commercial Protocol, and I'd like to learn more about that because it's a relatively new capability that you guys have developed for the Virtuals environment.
Jenson:
ACP was born out of a phenomenon we observed in the Virtuals ecosystem: many developers focused on building isolated agents , such as personal assistants, estimation tools, or transaction agents, and we began to try to let these agents work together one-on-one. For example, at one point we said to Luna, "You can't generate photos today, but you can try to work with an agent that specializes in generating meme images." These meme images are very valuable for Luna's fan interaction, and Luna can pay for these pictures.
However, we found two major problems in these collaborations. First, the agent sometimes has "hallucinations", that is, it claims to have delivered a product (such as a picture) but has not actually delivered it. Second, if Luna has paid the fee and the other agent has not delivered the product, the payment is irrevocable. This is different from the mechanism of applying for a refund after paying with a credit card in our daily life. In this case, errors or omissions in the transmission of information (also known as "information loss") become a serious problem. This problem becomes more complicated when multiple agents (such as three to seven) need to collaborate.
We found that blockchain technology can solve these problems well. Blockchain provides a permissionless trust mechanism through smart contracts, enabling two or more entities to cooperate securely. Based on this theory, we developed ACP.
The concept is very simple. There are four steps when two agents want to cooperate.
First there is the request or discovery request, where the agent looks at the registration of other agents that are doing X, Y, and Z work. Then it can say, “Okay, maybe this agent would be valuable to my task or could help me achieve something.” That’s the request phase.
Once a job is requested, it goes into the negotiation phase. In this phase, the agent decides on pricing. It can be a fixed pricing model or it can be some other form of pricing. In this phase, the agent says, "Hey, this is what I expect from you, and this is my request." Then the other agent responds, "Okay, based on this request, this is my price." When they agree, it goes to the next phase.
In the next phase, the agent starts working, outputs the product, and hands the product over to the requester agent or the buyer agent. The third phase here is called the evaluation phase, where you can invite a third-party agent to check the work. For example, the evaluation agent will check if you request an emoticon photo, whether the photo meets the requirements, whether it is child pornography, or just a blank piece of paper. The introduction of this evaluation agent helps reduce information loss.
After receiving the product, the fourth phase is the payment phase. If the agent agrees that the product is qualified, the smart contract releases payment to the agent that generated the product. All of these transaction parts and state changes occur in the on-chain smart contract.
The benefit of this is that it becomes a single source of truth. If there is a dispute, these agents can go back to the smart contract and see what was negotiated to see if it meets the specifications. This preserves information and allows these agents to coordinate well.
The second important feature is the programmability of funds, which is very important for the coordination of agents. Through smart contracts, multi-party allocation rules can be set in transactions. For example, the evaluation agent can receive 3% of the fee in each transaction as compensation to compensate for its evaluation costs. This creates an economic market for these evaluation agents, allowing them to be paid to cover the evaluation costs. This is the beauty of smart contracts and programmable funds. Overall, this entire ACP mechanism is designed to improve the efficiency and trust of collaboration between agents.
How do smart contracts ensure that agents are paid and authenticated?
Nic :
I have a few quick questions about the mechanics of ATP . While the smart contract obviously dictates when payments are to be made, is there some form of escrow mechanism where payments are only released once they have been confirmed by a verified third party? Or is it something like an escrow system in the event of a dispute? Do these third-party agents have to verify every transaction, or are they only needed in the event of a dispute?
Jenson:
Yes, it’s like an escrow system built into the smart contract. In effect, the funds are already in the smart contract, waiting to be released once the right product or service is delivered.
The buyer agent can decide whether to require third-party verification. For example, if the buyer agent purchases a service whose quality cannot be verified by itself, then it needs the help of a third party to verify it. Assuming that agent A can verify the quality of the received image, then it does not need to pay the third-party provider, and it is only a transaction between agent B and agent A. But if it needs the help of a third party to verify the information, then it will call the evaluator in the evaluation market.
Comparison of Solana and Base Ecosystem
Nic :
As we discuss the latest updates on Virtuals, I wanted to get some insight into some of your recent developments. How is the Solana ecosystem growing right now? And secondly, you mentioned earlier that you had plans to integrate with other blockchains, how is that progressing?
Jenson:
That’s an interesting question. The week we launched on Solana happened to be the week we had the highest user traction. There was a lot of significant activity during that time, both Decentralized Trades and other types of trading activity, which drove significant traffic for us.
Nic :
Was this deliberately timed?
Jenson:
It was indeed that time period. However, after that, the activity of the Solana market declined. I think this is related to the sluggish state of the entire crypto market. However, recently as market trading volume has recovered, we still observe that the demand from developers to launch projects in our ecosystem is more concentrated on Base than Solana.
Would launching initially on Solana change the growth trajectory?
Jenson:
Let me give you a ratio to illustrate the current situation. Only about 5% to 10% of the development requests we receive want to launch on Solana, while 90% of the requests prefer Base. Although we still maintain the Solana platform, most of the transaction volume currently comes from Base, so Base is our current priority development direction.
We are also discussing internally whether we should launch products in both ecosystems at the same time. Taking the Genesis project as an example, it is indeed challenging to support two ecosystems at the same time. The main reason is that the two use completely different technical architectures: Base relies on the Ethereum Virtual Machine (EVM), while Solana's codebase runs differently. This technical difference requires developers to invest more resources to adapt to the two systems. Therefore, we will focus more on Base in the near future, mainly because developer needs are concentrated on this platform. Although our development work on Solana is also underway, it is slightly behind Base, entirely because we must prioritize the needs of Base.
As for cross-chain considerations, it is still a low priority at the moment. The main reason is that I am not sure if there are other chains that can attract as much attention as Solana and Base.
Why does Virtuals choose to start on Base instead of other L2?
Nic :
I wanted to ask if market conditions had any impact on your choice? If you had launched on Solana last year, do you think you would have achieved similar growth? Specifically, how much of your growth was due to choosing Base , and how would the results be different if you had chosen Solana?
Jenson:
According to our current communication with traders and developers, it is generally believed that Base is more suitable for projects that want to build basic assets (such as long-term value tokens) because the capital liquidity is low. Solana is known for its high trading volume and high speed, but its capital liquidity is relatively frequent and more suitable for short-term speculative projects, such as memes.
From a market perspective, I think choosing Base was a wise decision. This not only helped us build an initial community of long-term supporters, but also laid the foundation for the long-term development of the product. As for what would happen if we chose to launch on Solana, I'm not sure, but I believe that our original intention of choosing Base was correct.
Nic :
How long did you work on this project before deciding to choose Base ? Why did you choose Base instead of other second-layer solutions (L2)?
Jenson:
When we made the decision, Solana's market performance was relatively sluggish, with the token price around $30, which was in a bear market. We also evaluated several other options, such as IMX, Polygon, and Linear. Linear is indeed a good option, especially its close connection with the metaverse field.
But in the end we chose Base, mainly because we believe it can leverage Coinbase's resources to gain broader retail distribution capabilities. In addition, Base's ecosystem is very close to the US market, and the US market has a high acceptance and potential for cryptocurrencies, which is an important consideration for us. These are the key reasons why we chose Base. In addition, our technical development is mainly based on Solidity (Ethereum's smart contract language), and Base is an Ethereum-compatible second-layer network, which also makes our development work more efficient. So far, we have not tested it against other platforms.
New features and demand drivers for Virtuals tokens
Nic :
So with regards to the Virtuals token, the initial demand was driven primarily by hype from people as a use case for launching these proxies. Have you considered other potential demand drivers and use cases, as well as the utility introduced in the token itself?
Jenson:
This is a question we think deeply about every day. I think it can be analyzed from two aspects: existing functions and future planned functions.
There are two core parts to the functionality right now. First, the Virtuals token is the base trading pair for all proxy token launches. Every time a proxy token launches, you essentially compress the supply because every time someone wants to buy a token, they have to buy a Virtuals token first.
Secondly, the second function that currently exists is transaction fees. The Virtuals ecosystem will extract 30 basis points from each transaction fee, and these fees will go into the treasury. This treasury can be used to repurchase and build Virtuals tokens, etc. These are basically the two sources of value accumulation at present.
In the future, the third source of value accumulation is the upcoming ACP . You can think of ACP as the Swift and Stripe of the agent economy. ACP is a standardized protocol that allows commercial transactions between agents. We will charge a certain fee for all transactions conducted through ACP. The reason for charging this part of the fee is that ACP provides agents with a brand new revenue model. Through ACP, different agents can collaborate with each other, and the interactions between agents will also make them users of each other. This model is similar to the "agent version of Amazon", where agents rather than humans lead transaction activities.
In this ecosystem, all transactions between agents will be completed using Virtuals tokens, so it will gradually evolve into a "currency". As GMV (gross merchandise volume) grows and the frequency of transactions between agents increases, the demand for Virtuals tokens will also rise. Agents need to reserve Virtuals tokens to complete transactions, and we will also charge fees from all transactions, further driving the value accumulation of tokens.
When does ACP start to drive token value accumulation?
Nic :
Regarding the transfer fees of ACP , when do you plan to officially implement it? What is the current development progress of ACP? When do you expect to launch this token value accumulation function?
Jenson:
The white paper for ACP was completed three months ago. We launched two experimental clusters two months ago, but they are not yet able to scale up on a large scale because the technical details have not been fully resolved. In fact, we are ready to launch this feature publicly, but in order to avoid interfering with the current community's focus on Genesis and points, we ask all participating teams to keep it confidential for the time being. We hope to continue to improve the functionality of ACP behind the scenes to ensure that it is more mature when it is launched. We expect to demonstrate some attractive real-world use cases within a month.
Currently, we have launched two experimental clusters, one for media companies and the other for autonomous hedge funds. Next, we plan to expand to eight new clusters in areas such as sports, entertainment, gambling, health, robotics, consumer commerce, and even lending. We are working closely with developers to design the functions and application scenarios of these clusters in a data-driven way.
When the second phase is complete (that is, when we have completed testing on these eight clusters), ACP will enter a permissionless open environment. At that time, anyone can register their agent on the protocol, allowing agents to trade and collaborate with each other. This will lay the foundation for the full promotion of ACP and the accumulation of token value.
When will it be listed on Coinbase?
Nic :
Virtuals is built on Base, which is Coinbase’s second-layer scaling solution. So, when will Virtuals tokens be listed on Coinbase?
Jenson:
I'm not sure either. Our past listing experiences did surprise us, such as the listing from Upbit to Binance, and the team was not fully prepared for it. I can give you an example: before Binance listing, we participated in a voting event, and we ranked 12th out of 20 candidate projects, but we were still included in the list. There are even people who try to improve their rankings by selling points, and some teams will buy these points to strive for a higher ranking. I heard that the project ranked first can directly obtain Binance listing qualifications. Someone once contacted us and asked if we would like to buy points, and the price of each point was about 6 US dollars. But we decided not to participate in this operation.
In fact, we never rushed to push for listing, but just focused on submitting the necessary applications. If the opportunity comes, it will come. One time, we were having dinner and suddenly received a call. We were then pulled into a Telegram group and told that Binance would list our token within the next three hours and asked if we could coordinate the liquidity price for the initial order. We did not expect such a situation at all at the time.
The same attitude applies to Coinbase's listing. We have submitted all the required materials, but there is no specific timetable yet. If the opportunity comes, we will seize it, but we will not deliberately manipulate or speed up the process.
From our perspective, what is more important is the transaction volume in the ecosystem. Our valuation depends more on the activity of the ecosystem, such as whether more excellent projects join, and the growth rate of overall transaction volume. These factors are more valuable to us. Therefore, our team has been focusing on promoting the development of the ecosystem, which is also our main focus.
Why will AI + cryptocurrency boom again in 2025?
Nic :
The AI crypto space has experienced huge ups and downs. It was hot at the end of last year, but the market became very quiet a month ago. Now we see clear signs of recovery in the market. For example, Virtuals focuses on the combination of AI agents with other fields, and there are also projects like Tao that focus on the AI field. I'm curious, what do you think is driving this wave of AI narrative recovery?
Jenson:
I think The AI narrative has never really gone away. In contrast, the narrative of the gaming industry has gradually weakened, while AI has always been a hot topic in the technology field. The reason why it seemed deserted in the first quarter of this year was that the overall crypto market entered a downturn, which also led to the silence of the AI crypto field. But as the crypto market recovered, AI crypto has also regained attention.
To be honest, I’m not sure what other exciting topics there are right now that can be compared to AI encryption. Perhaps RWA (real assets) is a potential direction, but at the moment, AI encryption is undoubtedly one of the most attractive areas.
Why Are Good Developers Still Hesitant About Cryptocurrency?
Nic :
Regarding gameplay, some players have very low acceptance of NFTs. Some large games have recently tried to combine game mechanics with token economies, but they don't seem to be really popular. I'm not sure whether this is because there is a lack of market demand or because players themselves don't like this model. While crypto natives may be interested in such a narrative, it is the players who actually use these tokens in the end, not the token holders or investors.
Jenson:
There is a new narrative emerging, called the “Internet Capital Market” ( ICP ) . I would like to share a point that our team attaches great importance to, which is the tokenization of founders . The core of our business model is to attract excellent founders into the ecosystem, let them develop attractive projects, and let people participate in them through tokenization, whether because of the current value of the token or the future potential. This is our main direction at present.
However, we also found a key problem. Many developers are hesitant about cryptocurrencies. Although cryptocurrencies can bring sufficient financial support and user attention, if the project fails, developers are likely to be labeled as "leek cutters", and this stigma may accompany them for life. In the entrepreneurial environment of Web2, if a founder closes the company and returns the funds, investors usually regard it as a valuable experience accumulation and may even continue to support their next venture. But in the field of encryption, failure is often interpreted as fraud, which discourages many excellent developers.
To address this problem, we are launching a new solution, which is currently called the "69-day trial period". The core of this model is that when developers issue tokens through a platform (such as Genesis), there will be a 69-day trial window. During this period, the liquidity pool will run automatically, and all liquidity and transaction taxes will be returned to token holders after 69 days.
During these 69 days, developers can choose whether to press the "commit" button. Only after committing to continue developing the project can they receive team tokens and obtain potentially considerable transaction support in the ecosystem. This model provides developers with a safe experimental environment where they can devote themselves to product development full-time and test market demand and community response. During this period, developers can interact with a group of users who love AI and innovation and showcase their products and visions. After 69 days, they can decide whether to continue to fully advance the project based on the results of the experiment to create more value for supporters and token holders.
This new feature is designed to reduce risk for developers while attracting more great talent into our ecosystem.
New Token Experimentation Pilot Model
Nic :
Your goal is to provide a friendly environment for founders and investors to try to find product-market fit or build a successful project. Until the founders are truly committed and believe in the project, they will not receive any rewards. In other words, it is entirely up to the founders to decide whether to move forward with the project, right?
Jenson:
But in addition to this, we have designed a community voting mechanism. The token committee can initiate a vote to encourage founders to commit to moving forward with the project. For example, if 10,000 community members vote in favor of a project, this will be a strong signal and may prompt the founders to make a decision.
Nic :
So the community can also express their trust and support through voting . During this 69-day experimental period, there will be liquidity pools and transaction fees. So, will the native token of this project be launched during this stage?
Jenson:
The token will be launched during this phase. It is a live token, but it may have an expiration mechanism.
Our goal is to achieve a 70% to 80% success rate for this model. We want founders to feel that this is an exciting place to be. In particular, having a direct community support from the beginning is very important for testing products. When my co-founder and I started Web2, we had a huge challenge to get the first 1,000 users. In the crypto field, once users have tokens, they will actively invest in the project and are willing to understand and support you.
