Rewriting Gaming Rules with AI: Michael Tong’s Journey from NetEase COO to All-in Web3

Michael Tong, the founder of Xterio, recently spoke with Wu Blockchain about his journey from the traditional gaming industry (NetEase, FunPlus, etc.) to establishing himself in the Web3 space. He shared his experiences in building within the Web3 ecosystem, detailing Xterio’s Layer 2 infrastructure, game development efforts, NFT marketplace initiatives, and integration of AI. Highlight projects like “Palio” and “Befriend” showcase the potential of AI in gaming. Michael also explored key differences between Web3 and Web2 gaming, the phenomenon of TON mini-games’ rise and decline, and the growing diversification in TGE (Token Generation Event) models. Lastly, he expressed his vision for the future of AI and how its synergy with gaming could bring more immersive and personalized experiences to users.

Read more：

https://medium.com/@XterioGames_29169/the-unreasonables-28958130237d

Introduction

Colin: Hello, everyone! Today, we’re here with Michael to discuss Web3 gaming. Michael, could you start by sharing your background and what you’re currently working on?

Michael: Sure, hi, everyone. I’m Michael, and I have over 20 years of experience in the gaming industry. I started at NetEase, where I participated in the launch of “Fantasy Westward Journey” about 20 years ago. I stayed with NetEase for a long time, leaving officially in 2014 as the COO. I also served on the board until 2021, making me the longest-serving board member after Ben.

After leaving NetEase, I ventured into investments and entrepreneurship. I became an investor in FunPlus and also started some smaller projects of my own, like female-oriented games. Now, I run a company called Xterio. Xterio is focused on Layer 2 ecosystems, and we’ve built a lot within this space. Our core lies in gaming, which includes games we develop in-house and those created by our partners. Additionally, we have our own launchpad, NFT offerings, and marketplace.

Our most popular game so far is an AI-native game called “Palio.” The beta version launched in April this year, about eight months ago, and it reached 8 million wallet addresses while generating over 1,000 BNB in revenue. More recently, we launched another AI game called “Befriend.” This game has even more unique active wallets (UAW) than Palio, and its daily active users (DAU) are currently around 2 million. While it hasn’t introduced any monetization features yet, it already has a massive player base. That’s a brief overview of what Xterio is working on.

Differences Between Web2 and Web3 Gaming: User Behavior and Business Models

Colin: You spent a significant part of your career at NetEase, including managing the iconic “Fantasy Westward Journey” IP. Transitioning from Web2 to Web3 gaming must have been quite a shift. What differences do you see between the two?

Michael: The differences are indeed substantial. At first, we didn’t fully understand the Web3 gaming market or its users. We lacked a sense of respect for what was essentially an entirely different industry. As a result, we didn’t always hit the mark in creating products that resonated with Web3 users, especially when it came to understanding their unique needs. But we’ve learned from our mistakes, gradually adjusted, and eventually started developing games tailored to different types of Web3 users.

Web2 users, on the other hand, are quite varied, and I like to think about them in terms of scenarios. For instance, in the early days of internet cafes, groups of people would gather to play competitive games. The atmosphere was electrifying — think “World of Warcraft,” where everyone would be shouting and collaborating in live raids. However, as gaming shifted to PCs and mobile devices, this kind of scenario faded. Mobile gamers, especially older generations, often play games more casually, mostly as a way to pass the time.

I have a famous friend who created the game “Black & White.” He once said that many people play match-three games not for any particular reason but simply out of habit — like having a glass of wine before bed to relax. This habit helps them wind down and fall asleep.

Web2 games span a wide variety of genres, but Web3 gaming has its own distinct characteristics. Many Web3 games start with “Play to Earn” models, attracting users by offering financial incentives. This is perfectly reasonable; after all, Web2 platforms also use similar methods, like ad-supported revenue models on novel platforms. The key challenge, however, is figuring out how to bring users into the ecosystem and effectively convert them into long-term players.

One critical aspect is transitioning users from “earning” to “spending.” I didn’t have much experience in this area during my Web2 days, but I’ve learned a great deal through Web3 projects. Take our game “Age of Dino” (AOD) as an example. Many people assume that Web3 gaming is just about using certain mechanics to direct users toward “farming” or earning rewards. But we found that many players are actually highly skilled and deeply engaged. Some are big investors who don’t just buy our NFTs but also actively play the game and become key players in alliances.

These players not only drive others to join the game but also enhance its competitive dynamics by defeating rival alliances. They often invest substantial amounts of money to upgrade their equipment, bases, and more. In my view, this showcases a critical opportunity for conversion in Web3 gaming. Such deep engagement is one of the most significant phenomena we’ve observed in this space.

The Rise and Challenges of TON Mini-Games

Colin: One of the hottest gaming trends this year has been the rise of TON mini-games. Were you following that trend? What’s your take on the TON gaming phenomenon, which seemed to rise suddenly but now feels like a flash in the pan?

Michael: Here’s how I see it. The key to TON’s success was its ability to convert Telegram users into Web3 users. I think this was a fascinating phenomenon. If you ask where the largest concentration of people in the industry resides, the answer is overwhelmingly Telegram. Discord is also significant, but Telegram undoubtedly has a larger user base.

Telegram’s user base includes many who are already familiar with Web3 or at least have some exposure to crypto. This makes it easier to convert them into genuine Web3 users, engaging them in activities like cryptocurrency trading. From this perspective, TON’s conversion funnel was highly successful. However, the challenge lies in whether these converted users are genuinely interested in gaming — particularly casual games. Retaining these users with engaging games is a significant challenge.

Right now, the buzz around TON seems to have cooled. I can’t say for certain why, as we haven’t done much in this area. However, I believe this kind of challenge is inevitable. Take STEPN as an example. Its success wasn’t just because it involved running shoes, but because it managed to break into a niche and achieve effective user conversion. It targeted a specific demographic and resonated deeply with them, which was the key to its success. It wasn’t the theme or the mechanics but the strategic conversion point that made it work. As of now, we haven’t seen another breakout project in Web3 gaming that replicates STEPN’s phenomenon.

This underscores how difficult it is to overcome these challenges. TON mini-games face similar issues. Of course, I hope they can quickly launch another project that breaks into a new niche, just like STEPN, haha.

Colin: STEPN really was a unique example. It combined elements like running, walking, and even shoe-gambling mechanics to create a one-of-a-kind phenomenon. But it’s unfortunate that they didn’t focus on building out the IP or pursuing long-term development. Instead, they veered into a Ponzi-like model, which eventually caused its decline. It’s a pity for the industry.

Michael: That’s true. They did try launching another project called Gas Hero, but it didn’t seem to gain much traction.

Colin: Yes, it feels like no one really paid much attention to that project.

Michael: It’s tough, for sure. This kind of challenge is massive, haha.

Xterio’s AI Exploration

Colin: You’ve recently made significant progress in AI-driven projects. Could you explain your approach and vision for integrating AI agents?

Michael: Sure, I can talk extensively about AI. Let me give you the full story. It all started with DeepMind’s AlphaGo in 2016 and 2017. I’ve always been a Go enthusiast — not particularly skilled, but deeply interested. I loved studying techniques from players like Lee Chang-ho and the life stories of legends like Go Seigen.

When AlphaGo defeated all the top human Go players, my admiration for human Go masters took a major hit. It was clear that AI had far surpassed human capabilities, leaving no room for resistance. Only Lee Sedol managed to win one game against AlphaGo, but that was it. Since then, I’ve been captivated by AI and its potential.

Later, I explored AI-based games, including some on ChatGPT 3, though their quality was still lacking. Everything changed with the advent of tools like MidJourney. I fell in love with it and shared many of my creations on Twitter. Between November and December 2022, after ChatGPT’s release, I began seriously considering how to integrate Web3 with AI.

The concept for Palio originated entirely from my personal vision, but bringing it to life required assembling a team that shared my passion for both Web3 and AI, along with the technical expertise to create games. This took a long time, but eventually, I built the right team, and we pushed the project forward.

In April of this year, we launched the beta version of Palio. We trained a custom image model for players, allowing them to generate pets with their unique styles and mint them as NFTs. We also partnered with Reka, whose core team includes members from Google’s AlphaGo project. When I first met them, I was blown away by their talent, and it’s been an incredibly exciting collaboration.

Palio’s first-generation Emotional Engine, a system designed to create emotional companions, was the result of this partnership. When it launched, it attracted over 8 million addresses, including 400,000 paying users. We designed monetization carefully, ensuring players wouldn’t have to spend excessively to dominate the game, like acquiring farming points on a massive scale.

Our goal is to turn Palio into an IP that can expand through other games. By January of next year, we plan to release a beta version that early NFT buyers can experience first. Meanwhile, we’re continuing to work with Reka to enhance the engine.

Another project, Befriend, offers a simpler but equally engaging concept. Its story revolves around AI gradually taking over humanity, and players must learn to communicate with AI and build rapport to survive. Every robot and scenario in the game is designed to be interesting, with player-AI interactions often leading to delightful surprises.

We hope Befriend evolves into a platform-level product that not only provides AI agents but also enables users to create backstories, events, and scenarios. This combination will generate even more narratives and gameplay opportunities.

Looking ahead, our ultimate goal is to build a truly platform-level product, which could take one to two years. In the long run, I believe AI’s potential goes beyond generating text, images, or music. It could become something akin to Doraemon — a mentor, a friend, and a super-robot capable of helping you with various tasks.

I see Xterio as having a real opportunity to pursue this vision. While training our models, we’re also exploring AI’s capabilities in vision and sound, such as recognizing users’ facial expressions and voice tones, and even displaying its own emotions and behaviors. It’s a distant yet worthwhile goal, and I’m confident that combining Web3 and AI can bring us closer to making it a reality.

The Potential of AI Agents in Gaming

Colin: Currently, we don’t see many integrations of AI agents with games. Most applications seem to focus on token issuance, such as memecoins, where automation is primarily used during the token launch process — for example, chatbots on Twitter. How are you considering integrating AI agents into gaming?

Michael: That’s an interesting question. Let me start by commenting on this phenomenon briefly. I’ve used my own wallet to experiment with some of these features, haha. I see two main pathways here.

The first involves using token launches to build a community. This approach allows some people to make money, which they can then reinvest into improving functionality.

The second involves applying AI agent capabilities directly to games. I’ve observed some projects that have started exploring this direction. For example, AI agents can be designed as non-playable characters (NPCs) within games. I think this approach is promising. By integrating AI agents into various games as NPCs, they can fulfill critical roles within the gameplay.

In the future, I believe AI agents could evolve to become a part of the game content itself. For instance, in games with insufficient content or a limited number of human players, AI agents can fill these gaps. Even if real players join later, they can immediately interact with AI, enhancing the game’s playability and interaction.

This is one way AI agents can be integrated into games, and it’s an effective way to quickly reduce the cost of content creation. Moreover, as these integrations reach a certain scale, qualitative changes might emerge. Applying AI agents to a wide range of scenarios could potentially transform gameplay, offering entirely new experiences. I think this potential is incredibly exciting.

The Growing Diversity of TGE Models

Colin: Your project seems to be gearing up for its TGE phase. How’s that progressing? Could you share some updates with the community or readers?

Michael: Sure! I think it’s going pretty well. Today is the 24th, and I’m sitting in a nice café picking out gifts for colleagues and partners. Haha, if things weren’t going smoothly, I’d probably be at home swamped with work.

Overall, things are progressing smoothly. When we’re ready to announce formally, the community will be the first to know. For now, our ongoing Befriend campaign is quite straightforward — we’re encouraging everyone to jump on board, haha.

Colin: Understood. Speaking of TGE, there have been significant changes in the overall model this year. For instance, Binance is no longer the sole option for listing. While participating in their Launch Pool requires substantial token payments, more projects are opting for platforms like BYBIT, Coinbase, and Upbit, and the results seem promising. Additionally, some projects, like Hyper, have bypassed centralized exchanges entirely, launching tokens independently and reaching impressive valuations. What’s your perspective on the diversification of TGE models?

Michael: I think diversification is undoubtedly beneficial for project teams because it provides more options. However, we also need to rely on data to evaluate specific outcomes rather than making judgments based on perceptions or superficial impressions.

For example, some might assume a particular exchange performs poorly in token listings, but analyzing the data could reveal a different story. Projects that launch through platforms like BYBIT and Upbit often perform well, leading people to think this pathway is particularly effective. However, if other projects taking the same route fail, the backlash may last only a few days before being forgotten.

In my view, decisions should be guided by objective data. Each method has its pros and cons, but no single strategy works universally. Even the same method can produce vastly different results depending on the project. Some succeed spectacularly, while others falter. Ultimately, it all comes down to analyzing each case individually.

Follow us
Twitter: https://twitter.com/WuBlockchain
Telegram: https://t.me/wublockchainenglish