Dialogue with MEXC Product Director Vivien: How to drive the next evolution when AI trading enters the "co-pilot" stage?

Compiled by: momo, ChainCatcher

Guest: Vivien, MEXC Product Director

At the Coinfest Asia 2026 event in Bali in August, the Alpha Arena S03 trading competition sponsored by MEXC Ventures had just concluded, with 20 traders from 12 countries battling it out in real-time, and a Japanese contestant clinching victory at the last moment.

Outside the competition arena, MEXC Product Director Vivien Lin discussed another, longer "competition": how exchanges can capture the influx of users and funds amid the wave of AI and RWA integration.

Before joining the crypto industry, Vivien Lin worked at Morgan Stanley in foreign exchange and complex derivatives. Her combined background in traditional finance and crypto allows her to better understand the significance of this integration and transformation.

She observed that with the introduction of perpetual contracts in traditional finance, the capital between the two markets has been interconnected, and user behavior is also converging. When BTC prices are sluggish, funds in the crypto circle flow into stocks and stock derivatives; when the market warms up, funds flow back into the crypto market. Changes on the product side are equally evident, as users inquire on MEXC AI about the differences between Nvidia and Intel, and hesitate over the authenticity behind tokenized stocks.

This change is also reshaping her product design logic. Vivien Lin initially considered creating two separate interfaces for two types of users but later realized they are more like different stages of the same group. Following this line of thought, MEXC has refined its user profiling, moving from a "one-size-fits-all" approach to a more layered "one-size-fits-one" model.

The explosive growth of AI is also rewriting users' trading habits and scenarios. In Vivien Lin's view, AI has transitioned from being a "helper" to a "co-pilot" for traders. MEXC's AI strategy goes beyond mere concepts, from acting as an information steward with MEXC AI to AI strategies capable of backtesting, and to AI search that directly accesses all functions with a single sentence; these features have quietly integrated into users' daily trading.

Recently, in an exclusive interview with ChainCatcher, Vivien Lin elaborated on how her traditional finance background shapes her product perspective, how to layer services for two types of users, the logic of invisible product design and AI tools, and the changes in exchange product forms over the next three years.

1. From Investment Banking to the Crypto World: Changes in Product Perspective

1. ChainCatcher: You previously worked in foreign exchange and derivatives at large investment banks like Morgan Stanley. How has this traditional finance background shaped your product perspective? Which logics from traditional finance can be seamlessly applied in the crypto world, and which must be discarded?

Vivien Lin: Traditional finance provided me with excellent financial training, making me more aware of risks. I traded complex derivatives and had to monitor Greek letter risks daily, many of which lacked ready market hedging tools, making me very sensitive to even small data fluctuations. This sensitivity helps me notice minor points when developing trading products.

It also instilled a sense of reverence for the market and regulation. Standing at the intersection of traditional finance and crypto, familiarity with regulation offers unique product insights. When designing compliant products, I better understand what regulators focus on and how to maintain user experience while meeting regulatory requirements, creating products that are both compliant and user-friendly.

However, the pace of traditional finance does not work in the crypto space. Developing a new product there takes 6 months to a year, allowing time to ponder user needs and business models; whereas the crypto world operates 24/7, forcing you to quickly judge based on intuition about the market, user psychology, and the product itself, while rapidly coordinating resources within and outside the company to push products out quickly.

2. ChainCatcher: In the past year or two, many traditional financial platforms have expanded into crypto, with "super accounts" becoming a trend. In contrast, what are the shortcomings and barriers of crypto-native exchanges?

Vivien Lin: First, the shortcomings. Early on, crypto market product design prioritized speed and lacked regional regulatory requirements, essentially offering a single product globally. Now, regulatory requirements vary by region, and users are more diverse, including novices, high-net-worth individuals, professional institutions, arbitrage funds, and those coming from traditional finance who are unfamiliar with crypto but have strong learning abilities. Whether we can solidify the infrastructure while accommodating diverse user needs and regional regulations will be a core challenge for exchanges to build their competitive moat in the next 1 to 2 years.

Regarding barriers, crypto-native exchanges excel in retail products; this market started with retail users, and institutions have been slow to enter. Traditional financial institutions understand finance and risk better, but they lack the understanding of retail users compared to native exchanges; native exchanges iterate quickly based on deep user understanding, while what they need to learn from traditional finance is professionalism.

2. User Behavior in Crypto and Traditional Finance is Converging

3. ChainCatcher: There are significant differences in trading interfaces and functional needs between crypto-native users and traditional finance users. How does MEXC balance the experience conflicts of these two types of users when creating a one-stop platform covering stocks and ETFs?

Vivien Lin: The primary principle of the product team is to provide users with a smooth experience. Based on that, the key is to observe what users truly need. I often cite two examples.

One is copy trading, a very crypto-native product. For seasoned crypto users, it’s an old play, but for traditional finance users, it’s a new tool; they need to find credible experts to guide their trading. So, on one hand, we provide selection criteria to help users judge what kind of traders are considered experts; on the other hand, we introduce more assets to give these traders more room to operate. This offers a fresh experience for traditional finance users.

The second is our newly launched RealStock, which caters to the demands of traditional finance users. Crypto users care about leverage and smooth operations, while traditional finance users primarily care about whether they actually own the stock when they buy it. With many tokenized stocks on the market, they always have a question mark in their minds, which can hinder their experience. After realizing this, we discussed it repeatedly and ultimately decided to directly connect with traditional brokers to gain real stock exposure. This was a technical path chosen to address users' core concerns.

4. ChainCatcher: Real stocks and traditional finance perpetual contracts are key directions MEXC is currently promoting. Based on current user data and product feedback, what are the behavioral differences between the two types of users regarding these products? What optimizations has the product team made in response to these differences?

Vivien Lin: Research suggests that the overlap between traditional finance and crypto users exceeds 60%, but our own experience hasn't reached that number yet.

However, after the introduction of traditional finance perpetual contracts, the capital between the two markets has indeed been interconnected. Recently, when BTC was sluggish, a significant amount of crypto funds flowed into stock tokens and stock perpetuals; now that the crypto market is picking up, funds are flowing back from stock positions to mainstream crypto assets.

Once the capital is interconnected, behaviors are also converging. Two years ago, we began promoting the integration of traditional finance users and products with the crypto space. At that time, traditional finance users criticized crypto mainly for its lack of transparency and being concept-heavy. Many platforms had launched index products or stock ETF-type products like QQQ, but they did not trust the authenticity of the underlying trades and merely observed from the sidelines. Now, more and more traditional users actively seek to understand how perpetual contracts settle, and with our AI-assisted learning tools, the understanding cost is not high. Some of my friends who are traders even say this mechanism directly solves the cumbersome issue of leveraging in traditional finance.

We once considered creating two separate interfaces for the two types of users. Later, we found that users' thinking actually follows the attention given to assets. When the AI sector is booming, many crypto users also ask on MEXC AI about the differences between Nvidia and Intel. Thus, the product logic shifted from serving two groups to serving different layers of the same group, categorized by maturity such as novice and experienced traders, rather than by asset type.

5. ChainCatcher: MEXC now covers over 170 markets. How do you design layered experiences and growth systems for different levels of users?

Vivien Lin: There are parts that users can perceive and parts they cannot.

The parts that cannot be perceived rely on more powerful analytical tools and big data technology. As soon as users enter the platform, we can determine their entry point, whether they are closer to traditional finance profiles or risk-averse, whether they are trading-oriented or deposit-oriented. In the early days, user profiling was very coarse, essentially divided into three categories: novices, institutions, and VIPs. In fact, many VIPs' operating habits are already very close to those of institutions, but at that time, data and analytical tools were insufficient, and the entire industry had a coarse granularity regarding users. Now, profiling can be detailed into dozens of categories; for example, a conservative user who only trades BTC and Ethereum and cannot tolerate large drawdowns would be recommended AI strategies focused on mainstream trends or reversal signals, with small drawdowns and backtesting support.

The parts that users can perceive involve the new VVIP system. Traditional VIPs are generally divided into levels 1 to 9 based on deposit amounts and trading volumes, while VVIP innovatively introduces M Score. It not only recognizes users based on assets but also converts all behaviors after users enter the platform into scores, resembling a loyalty card. Daily logins, participation in activities, and trying new products all accumulate loyalty. The more loyal users are, the more opportunities we provide for new product experiences, high-yield financial products, and higher cashback ratios for MEXC cards. The entire VVIP system and user profiling system are linked to M Score, with relatively transparent rules, making it one of the more advanced and user-perceptible tiered products in the market.

6. ChainCatcher: Can you share a specific case of localized feedback? How has it driven the iteration of core functions?

Vivien Lin: In the past three months, we have invested a lot of energy into something users may not necessarily perceive: optimizing internal operational tools. Previously, operational colleagues could only configure activities based on personal understanding, leading to significant generational differences; a 10-year veteran and a new operator, or someone from the 80s and someone from the 00s, have completely different perceptions of users. Covering over 170 countries and regions with users of different ages and risk preferences makes precise operations by humans nearly impossible.

We utilized AI and big data to refine user profiling, and through data engineering, we established a set of atomized internal products that allow operational colleagues to quickly launch and configure activities.

Now, for different countries and partners, configuring an event that aligns with current hot topics and community characteristics takes less than an hour. Previously, writing copy, producing images, and configuring multiple languages took two to three days. For example, when the market surged yesterday, the entire process from idea to launch used to take at least two to three days; now it can be completed within half an hour, allowing users to participate in the market at the right moment, complemented by promotional activities to enjoy the rise.

7. ChainCatcher: Besides what users can see, what other invisible product designs have played a key role in MEXC's retention and conversion?

Vivien Lin: We have done a lot beneath the surface. First, there is technical optimization; we are undergoing a major system upgrade. After the upgrade, users will directly feel two things: the app opens and switches faster and smoother. Sensitive users may say it’s a bit faster, while most users just feel more comfortable, unable to pinpoint what’s better, but for the underlying infrastructure, this is a transformative effect. The new infrastructure can accommodate a larger transaction volume, ensuring the system does not crash during extreme market conditions, and also supports our future exploration of lending and options products. This is a sense of psychological security provided by the product research and development team to users.

Another aspect in the account system and trading requirements may not be perceived by users. For example, AI strategies; users see that even in a deep bear market, the strategies you promote still yield decent returns. Where do these strategies come from? Behind them is a product team of four or five people who spent three months studying the trading strategies and styles of investment masters throughout history, and then localized them based on the changes and volatility characteristics of cryptocurrency assets, resulting in the polished product.

III. How will trading scenarios evolve in the AI era?

8. ChainCatcher: MEXC has launched several AI tools. Based on data and user feedback, what actual transformations have they brought, and have they increased trading frequency?

Vivien Lin: AI products can be divided into several categories. The ones that users can perceive and have clear intentions are MEXC AI and AI strategies. MEXC AI is more information-oriented, like an information steward, gathering global news and the technical indicators you need to see. Its design purpose is very clear; we look at how many users are actually using it. Users vote with their feet; if they feel you’ve done well, they will come back every day to ask and check. If they treat it as part of their daily information source, we have more opportunities to recommend current popular cryptocurrencies and trading opportunities to them, making it more likely they will see them and ultimately convert.

The intention of AI strategies is more direct; users basically need to reach the order placement step to build a trading system. We mainly look at two indicators: first is the repurchase rate; users will repeatedly use it only if they are happy with it; second is how many users actually build strategies and run backtests. This also deepens our understanding of platform users—how many are more professional and how many come in, find it difficult, and leave. The latter needs to distinguish whether it’s a product issue or if they were never the target users.

There are also efficiency-oriented products, such as AI news, AI search, and the command line interface we are about to launch, which are part of the early layout for the next generation of product experiences. Taking AI search as an example, it solves the end-to-end jump problem. Previously, if you wanted to find a financial product from the homepage, you had to first find the button to enter the financial section, then scroll through a long list; if you wanted to find niche assets, you would scroll for a long time and might miss similar names; on mobile, the page is limited, and after a few pages, you might not want to scroll anymore, with many useful products actually hidden behind. Now, if you enter BTC in the search bar, it will directly give you the three most important news items, the current price chart, technical indicators, ongoing activities, and whether there are financial products or income-generating products related to BTC—all in a very small search box, gathering extremely rich materials. This cannot be achieved through manual hard coding; now, the underlying components, interface calls, and content presentation are, to some extent, intelligently filtered by AI. When users search for a term, they can see almost everything related to it on this platform.

9. ChainCatcher: Based on the development of AI, can you share a representative future trading scenario?

Vivien Lin: I understand the development of AI in three stages.

The first stage is the assistant stage, where you ask a question, and it answers a question.

We are now entering the second stage, which is the copilot stage, a phase that follows closely. Under a prior authorization agreement, AI knows what I am doing. It will observe me; for example, if it finds that I tend to be impulsive during a market crash, the next time I feel impulsive, it will remind me: instead of buying a high-leverage product here, why not buy spot, or I have a high-interest deposit here; would you like to secure a guaranteed return first?

Looking further ahead, it will be the intelligence of operations, bringing a completely new experience. Now, if we want to find a function, we have to open the app and scroll to the corresponding page; in the future, there will be a unified entry point, such as embedding MEXC’s assistant into the phone system, connecting to Siri. You can directly tell Siri: "Today BTC looks like it’s rising well; can you help me make 1000 dollars?" Siri will call our command line interface, connecting to all backend services, including backtesting, and then judging whether it’s possible to make that 1000 dollars today based on my current asset reserves, and then execute it directly, or if it’s uncertain, give me three options to choose from.

This will approach the highly intelligent world depicted in "The Matrix." The technical barrier mainly lies in computing power. What application layer companies can do is make the calls more efficient, cost-effective, and precise. We cannot solve hardware and infrastructure issues, but once the infrastructure problems are resolved, the product experience will undergo a disruptive change, which I believe we will see within one to two years.

10. ChainCatcher: Looking ahead to the next year, what are MEXC's core plans for AI exploration and functional upgrades?

Vivien Lin: From a product perspective, AI technology has penetrated every aspect of our products and research and development. However, the externally perceivable products that users can sense still need to return to the first principles: being user-centric. We do not need to showcase to users that this platform uses AI the most; rather, we hope to subtly integrate our understanding of AI and users into every small function and every piece of information. You may not realize that you are using many AI functions; you just feel that this platform understands me better, which is the picture we want to create.

To reach that picture, we need to do a lot of foundational work first. First, we need AI to better understand assets, which requires continuous learning of the market; second, we need AI to better understand users, which relies on user profiling and big data. AI solves analytical problems; it cannot handle tasks like data storage, organization, and standardization, which require a lot of manpower to lay a solid foundation for it, allowing AI to leverage its strengths.

The product direction will also follow the market. Recently, the market has been relatively sluggish, and trend-based products have not been well received by users, which is also the reason we are developing AI strategies. AI strategies include both trend-based strategies and strategies that profit from price differences during bottom consolidation periods. It is precisely because of the preliminary technical and data groundwork that the product has been relatively smoothly pushed out.

11. ChainCatcher: Looking ahead to the next three years, as Crypto integrates with traditional finance, what fundamental changes will occur in the product forms of exchanges? What next-generation trading scenarios is the MEXC product team preparing for in advance?

Vivien Lin: Our biggest core is to serve users who come from traditional finance to the cryptocurrency space. Customer service, business development, partners, and product strategy will now spend more effort understanding what assets these users want, what tools they need, and whether our technical infrastructure can be upgraded to provide them with a speed and variety of experience comparable to traditional finance. This is the most important strategy from this year to next.

Under this strategy, the operations team needs to look at traditional financial assets, and the business development team needs to strive for traditional financial users and channels. For the product team, a major task is to fill in those products that are very handy in traditional finance but are not yet available in the cryptocurrency space. The major categories are basically not lacking; what’s missing are the details, such as iceberg orders, which are a very popular order type in traditional markets, but some exchanges in the cryptocurrency space have them while others do not. We need to quickly fill in these small points so that when users who truly engage in traditional finance come to the cryptocurrency space, we can meet their needs and serve them well.