Xterio founder Michael Tong accepted an interview with Wu Shuo, sharing his journey from the traditional gaming sector (NetEase, FunPlus, etc.) to Web3 and the directions he is exploring in the Web3 ecosystem. Michael detailed his company's layout in the Layer 2 ecosystem, including game development, NFT marketplace, and integration with AI. Key projects like 'Palio' and 'Befriend' showcase the application potential of AI technology in gaming, while discussing the key differences between Web3 and Web2 games, the phenomenal boom and decline of TON mini-games, and the diversification trend of TGE (Token Generation Event) models. Michael also shared his vision for the future of AI, hoping to create a more immersive and personalized user experience by combining AI and gaming.

Read more:

https://www.odaily.news/post/5200632

Opening introduction

Colin: Hello everyone, today we are here with Michael to discuss the gaming aspects related to Web3. First of all, Michael, could you please briefly introduce your resume and what you are currently doing?

Michael: Hello everyone, I am Michael. I have about 20 years of experience in the gaming industry. I started at NetEase, where I participated in the release of (Fantasy Westward Journey), which was about 20 years ago. I spent a long time at NetEase, until I officially left in 2014, when I was COO. In fact, I was on the board of NetEase until 2021, so among the board members, I have been there the second longest time, aside from Ben.

After leaving NetEase, I made some investments and started my own ventures, and I also became an investor in FunPlus. During this time, I tried running some small companies of my own, such as women's gaming projects, etc. Now I am running a company called Xterio. Xterio is a company dedicated to Layer 2, and our Layer 2 ecosystem contains a lot of content. Our core is gaming, including games we developed ourselves and games from partners. At the same time, we also have our own launchpad, NFT, and marketplace.

The game with the most users right now is an AI-native game called Palio. Palio launched its beta version in April of this year, which was eight months ago. At that time, it already had 8 million wallet addresses, and generated more than 1,000 BNB in revenue. Recently, we launched another AI game called Befriend, which has even more UAW (Unique Active Wallets) than Palio, with about 2 million DAU (Daily Active Users) currently. This game has not opened any monetization points yet, so revenue has not been calculated, but the number of players is very high. This is a brief overview of Xterio.

Differences in user attributes and business models between Web2 and Web3 games

Colin: So you have been in the gaming industry for a long time at NetEase, including being the head of a classic IP like (Fantasy Westward Journey), then transitioning from Web2 games to Web3 games. Wouldn't there be a significant contrast? What do you think are the differences between them?

Michael: Well, I think the difference is quite large. At the very beginning, the understanding of the user market for Web3 games may not have been very deep, and the respect for the barrier between industries was not enough, so at the start, we didn't fully realize whether what we needed to do matched the users accurately, especially in understanding the needs of Web3 users. But we learned through mistakes, gradually adjusted, and slowly managed to create games targeted at different Web3 users.

Web2 users actually have many classifications, and I like to describe these users with different scenarios. For example, in the era of internet cafes, a group of people might gather to play the same competitive game; that atmosphere was very exciting, and everyone would cheer for each other, such as when playing (World of Warcraft), where everyone would do dungeons together, and the on-site interaction was very strong. However, later, in the PC and mobile game era, that scene-based atmosphere weakened, and mobile game players leaned more towards casual gaming, especially the older generation, who primarily play mobile games to pass the time.

I have a famous friend who is the game producer of (Black and White), and he once said that many people play match-3 games without any special reason; it's just a habit, like having a glass of red wine before bed. Through this habit, they can relax and fall asleep.

There are many types of games in Web2, while Web3 games have their unique characteristics. Many Web3 games initially adopt some 'Play to Earn' models to attract users, allowing players to participate through some methods of making money. I think this is quite normal; in fact, Web2 also has similar models, such as novel platforms attracting users through advertising revenue. However, how to attract these users and achieve effective conversion is a very important issue.

The shift from 'making money' to 'consumption' is also a key exploration point. In Web2 games, I may have somewhat insufficient experience in this regard, but I have learned a lot in the process of Web3. Taking our practice in the AOD (Age of Dino) game as an example, many people think making Web3 games is just about using certain mechanisms to guide users to farming (yield generation). However, we found that many players actually know how to play games very well; they might be major players, not only buying our NFTs but also becoming key players in alliances after entering the game.

These players not only drive other players to play games together but also enhance the competitiveness of the game by defeating other alliances. They are willing to invest substantial amounts of money to upgrade their equipment, bases, etc. I think this reflects, to some extent, the conversion opportunities in Web3 games. This deep engagement is an important phenomenon we observed in this process.

The rise of TON mini-games and the challenges of continued development

Colin: Actually, the current hot game trend is still the previous TON mini-games. I wonder if you were paying attention at that time, and how do you view the overall gaming phenomenon of TON? It feels like it has gone from sudden popularity to a bit of a flash in the pan now.

Michael: Well, this is how I see it. The key point of TON is that it successfully converted users within Telegram into Web3 users. I think this is a very interesting phenomenon. Because if you ask which community or platform has the most people engaged in this industry, the answer is mostly Telegram. Of course, Discord can also be mentioned, but Telegram definitely has more users.

Telegram’s user base itself has many who already understand Web3 or at least have some exposure to cryptocurrency. This makes their user base very easy to convert into real Web3 users, such as starting to participate in virtual currency trading, etc. From this perspective, TON's conversion path has been very successful. But the question is, whether these converted users are truly interested in games, especially casual games? This involves what kind of games can retain these users, which is clearly a very challenging issue.

Currently, it seems that the popularity of TON has cooled down, but I am not sure about the specific answer. After all, we haven't done much in this field. However, I believe this challenge is inevitable. Taking STEPN as an example, its success is not just because it is a pair of running shoes, but because it achieved a breakout phenomenon. It reached a conversion in a specific area, which is key. But its success is not due to using a specific theme or model; instead, it found specific conversion points. As of now, we have not seen another explosive project that can break through like STEPN.

This also indicates that such challenges are very difficult. TON's mini-games will similarly face comparable challenges. Of course, I also hope they can quickly launch another breakout project like STEPN, haha.

Colin: Yes, STEPN is indeed a relatively special example. It combines running, walking, and some external elements like betting shoes very well, creating an accidental phenomenon. However, it is regrettable that they did not continue along the path of IP or long-term development, but instead took a route leaning towards a Ponzi model, leading to a gradual decline in later development. This is indeed a regrettable phenomenon in the industry.

Michael: Yes, however, they later launched a project called Gas Hero, but it seems to have not attracted much attention.

Colin: Yes, it feels like everyone hasn't paid much attention to this project.

Michael: Well, it's quite difficult. I think this is indeed a big challenge, haha.

Xterio's exploration in the field of AI

Colin: Regarding your recent progress, I see that your project may develop towards AI agents. Can you elaborate on this?

Michael: Sure, actually I can say a lot about AI. Let me share the longest version. It can actually be traced back to DeepMind's launch of AlphaGo in 2016 and 2017. At that time, AlphaGo defeated Go players, and I am a Go enthusiast myself, although my skill is average, I really enjoy researching, such as watching Lee Chang-ho's moves and Wu Qingyuan's life story, etc.

But after AlphaGo defeated all top Go players, my admiration for Go masters completely dropped, feeling that AI is much stronger than humans, with no power to fight back. Only Lee Sedol is the only human who has won a game against AI, but that hasn't happened again. Since then, I began to pay attention to AI-related matters.

Later, I also played some AI-based games, such as those on ChatGPT 3, but the quality of those games was not high at that time. It wasn't until the emergence of MidJourney that I really liked it and shared many of my generated works on Twitter. From November to December 2022, with the launch of ChatGPT, I began to seriously consider how to combine Web3 and AI.

The initial concept of Palio was entirely my personal idea, but to realize it, I needed to find partners who were interested in both Web3 and AI, and who had the capability to create games. This took me a long time, and eventually, I formed a team to advance this project together.

In April of this year, we launched the Beta version of Palio. We trained a dedicated image model for players, allowing them to generate pets in their own style and mint them as NFTs. We also collaborated with our partner Reka, whose core team members come from the Google AlphaGo project. When I first met him, I thought he was amazing, and this collaboration excites me very much.

The first generation of Palio's 'Emotional Engine' was completed through this collaboration. This engine is an emotional companion system that attracted over 8 million addresses at launch, including 400,000 paid users. We designed monetization points, but we won't let players spend a lot of money to dominate the game, such as acquiring farming points on a large scale.

We hope that Palio can become an IP, which we can further spread through different games. By January of this year, we plan to release a Beta version, allowing early NFT purchasers to have priority access. Meanwhile, we are also continuing to collaborate with Reka to push the iteration of the engine.

Another project, Befriend, is a simpler but equally interesting gameplay. Its background story is about how AI gradually dominates humans, and players need to learn how to communicate with AI and improve their favorability to continue surviving. Each robot and scene in this game is very interesting, and interactions between players and AI sometimes bring unexpected surprises.

We hope that Befriend can become a platform-level product, providing not only AI agents but also creating background stories, events, and scenarios for users, which will bring more stories and gameplay.

In the future, our goal is to build a truly platform-level product, which may take one to two years. In the long term, I believe the potential of AI is not just in text, images, or music generation, but to become your mentor, friend, and even a super robot that helps you complete various tasks, like Doraemon.

I believe Xterio has the opportunity to work towards this direction. While training the models, we are also exploring AI's visual and auditory capabilities, including recognizing users' expressions and voices, and even having its own emotions and behaviors. This is a distant but worthwhile vision, and I believe that the combination of Web3 and AI can bring this goal closer to reality.

The development potential of AI agents combined with games

Colin: However, we see that the combination of AI agents and games does not seem to be particularly abundant at the moment; most applications are mainly focused on token issuance, right? For example, memecoins, their functionality is currently quite single, maybe implementing some automation during the token issuance process, like chatbots on Twitter. How are you considering integrating AI agents with games?

Michael: Well, I think this is the case. Let me comment briefly on this phenomenon; in fact, I have tried these functions with my own wallet, haha. I think there are probably two paths here. First, some projects establish communities through token issuance; this indeed allows some people to make money, and then they use these resources to improve functionalities.

The second is to directly apply the functionality of AI agents into games. In fact, I have seen some projects that have attempted this direction to some extent. Specifically, AI agents can be designed as NPCs within games. I think this approach is quite good. AI agents can enter different games in this way and take on the role of NPCs.

In the future, I believe AI agents may further serve as part of the game content. For example, in cases where some game content is insufficient and the number of real players is limited, AI agents can fill these gaps first. Even if real players come in later, they can play directly with AI, which to some extent enhances the playability and interactivity of the game.

This is one way of integration and an effective method to quickly reduce the cost of game content production. Of course, sometimes quantitative changes can lead to qualitative changes. When this combination is applied to enough scenarios, it may change the entire gameplay and experience, bringing a completely new feeling. I think this possibility is very much worth looking forward to.

The diversification trend of the TGE model

Colin: It seems like your project is about to start TGE-related work soon. How is that work progressing? Is it going smoothly? Can you briefly introduce it to the community or readers?

Michael: Sure, I think it is still going relatively smoothly. Today is the 24th, and I can still sit in a nice café selecting gifts for colleagues and partners. Haha, if it weren't going smoothly, I would probably be at home busy with work right now.

So overall, it is still quite smooth. We plan to release information to the community as soon as we officially announce. In fact, our current Befriend activity is quite direct; we just hope everyone will hop on board quickly, haha.

Colin: Understood. Speaking of TGE, there has indeed been a significant change in the overall model this year. For example, Binance is no longer the only choice for listing coins, but if you want to be listed on their Launch Pool, you may need to pay a lot of coins. Relatively speaking, more people are starting to strive for a combination route like BYBIT, Coinbase, Upbit, and it seems to be working quite well. In addition, projects like Hyper, which completely rely on self-issued tokens without centralized exchanges, have directly surged to a very high market value. What are your personal or team views on the current diversification of the TGE model?

Michael: I believe, first of all, different approaches are definitely more beneficial for project parties, because it means more choices. Secondly, we also need to understand specific performances and effects through data, rather than relying solely on feelings or superficial judgments.

For example, sometimes people may think that the listing performance of certain exchanges is poor, but when you look at the specific data, you may find that the situation is not like that. For instance, some projects perform very well after being listed on BYBIT and Upbit, leading people to think that this path is particularly excellent. However, if other projects on the same path fail, people might complain for two days and then forget about it.

Therefore, my point of view is that we must rely on objective data for judgment. Different methods have their pros and cons, but they may not apply to all projects. Even the same method can sometimes yield varying results. Some projects perform excellently, while others may fail; this all needs to be analyzed based on specific situations.