At Coinfest Asia 2026 in Bali in August, MEXC Ventures had just wrapped up its Alpha Arena S03 trading competition. The event featured 20 traders from 12 countries competing live, with a Japanese participant taking the title after a late comeback.
Outside the arena, MEXC Product Director Vivien Lin was focused on a different contest: how exchanges should absorb incoming users and capital as AI and RWA trends begin to intersect.
Before entering crypto, Lin worked at Morgan Stanley on foreign exchange and complex derivatives. She said that mix of traditional finance and crypto experience has shaped how she looks at risk, regulation, and product design at a time when the two markets are moving closer together.
According to Lin, the introduction of traditional finance perpetual contracts has already opened a channel for capital to move between markets, and user behavior is starting to align as well. When BTC was weak, crypto capital shifted into equities and equity-linked derivatives. As crypto markets improved, some of that money rotated back into major digital assets. She said those changes are visible on the product side too: users ask questions on MEXC AI about the difference between Nvidia and Intel, while others still hesitate over whether tokenized stocks are backed by something real.
That shift has changed her product logic. Lin said she once considered building two separate interfaces for two different user groups, then concluded they looked more like the same people at different stages. MEXC has since moved from a broad one-size-fits-all model toward a more segmented structure with more than a dozen user profiles.
She also said AI’s role in trading has changed. In her view, it has moved beyond an “assistant” function and into a “copilot” phase. MEXC’s current AI stack, she said, is not just a concept: it includes MEXC AI as an information manager, AI strategy tools that can run backtests, and AI search that lets users reach platform functions with a simple prompt.
How traditional finance shaped her product approach
Asked how her earlier years at large investment banks influenced her product thinking, Lin said traditional finance gave her strong technical training and a much sharper sense of risk. She used to trade complex derivatives and track Greek exposure every day, including positions that did not have ready-made hedging instruments in the market. That made her highly sensitive to even very small moves in the data, and she said the habit still carries over into how she evaluates trading products now.
She added that traditional finance also taught her to respect both markets and regulators. At a moment when traditional finance and crypto are beginning to merge, she said regulatory familiarity creates a different kind of product insight. It helps her understand what regulators are focused on and how to preserve usability while still meeting compliance requirements.
What does not transfer cleanly is the pace. Lin said a new product in traditional finance can take six months to a year to develop, with time to test user demand and business logic. Crypto does not work that way. The market runs 24/7, and teams are forced to make quick calls based on market feel, user psychology, and product intuition, then align internal and external resources fast enough to ship and iterate.
Where crypto-native exchanges are weak, and where they still hold an edge
As more traditional finance platforms add crypto products and the idea of a “super account” gains traction, Lin said one weakness of crypto-native exchanges comes from their early design logic. Products were built for speed, often as a single global offering, because regional regulation was less differentiated. That setup is much harder to sustain now.
She said exchanges today face a more fragmented reality: different jurisdictions have different regulatory requirements, and the user base is much more varied, spanning beginners, high-net-worth individuals, institutions, arbitrage capital, and users coming from traditional finance who may not know crypto well but learn quickly. Over the next one to two years, she said, the ability to build strong infrastructure while handling both regulatory diversity and diverse user demand will be a core test of whether an exchange can build a moat.
Still, Lin argued that crypto-native venues retain a real advantage in retail products. The market was built by retail traders, while institutions entered much later. Traditional finance institutions understand finance and risk better, she said, but they do not understand retail users as deeply as native exchanges do. For crypto-native platforms, the area where they still need to catch up is professionalism.
User behavior is converging across crypto and traditional finance
On how MEXC balances the conflicting needs of crypto-native and traditional finance users on a platform that includes stocks and ETFs, Lin said the first rule is to keep the experience smooth. After that, the real task is to understand what users actually want.
She pointed to copy trading as one example. It is a very crypto-native product for long-time digital asset users, but for traditional finance users it can feel like a new tool because they are looking for credible traders to follow. MEXC’s response, she said, has been to provide screening standards that help users judge which lead traders are worth trusting, while also expanding the asset universe available to them.
Her second example was MEXC’s newly launched RealStock product. Lin said this was designed around a core concern from traditional finance users. Crypto users tend to care first about leverage and execution smoothness. Traditional finance users ask something else first: if they buy a stock, do they actually get real stock exposure? She said tokenized stocks in the market still leave many users with an unresolved question about whether the underlying exposure is real. After internal discussion, MEXC decided to connect directly to traditional brokers in order to obtain real stock exposure. That, she said, was the technical path chosen to address the user’s biggest concern.
Lin said RealStock and traditional finance perpetual contracts are both major areas of product push right now. She noted that some research puts the overlap between traditional finance and crypto users at more than 60%, though she said MEXC’s own reading of its user base is lower than that.
Even so, she said the launch of traditional finance perpetuals has clearly connected capital across the two markets. During the recent period of weak BTC performance, crypto money moved heavily into stock tokens and stock perpetuals. As crypto prices recovered over the past few days, funds rotated back out of equity positions and into major crypto assets.
Once the capital path opened, she said, user behavior began to converge too. MEXC started promoting product and user integration across traditional finance and crypto about two years ago. At that time, many traditional finance users criticized crypto as opaque and overly conceptual. Some platforms launched products tied to indexes such as QQQ or stock ETF-style exposures, but many users did not believe the underlying trades were real, so they stayed on the sidelines.
That is changing, Lin said. More users with a traditional finance bias are now actively trying to understand how perpetual contracts settle, and MEXC’s AI learning tools help reduce the cost of understanding. She added that some friends who work as traders have told her the mechanism solves one long-standing frustration in traditional finance: using leverage there can be cumbersome.
Lin said the team seriously considered building two different interfaces for two user groups. Later, it noticed that user attention followed whichever asset class was hot. When AI-related names surged, many crypto-native users were asking MEXC AI about the difference between Nvidia and Intel. That changed the design framework. Instead of serving two isolated groups, MEXC now treats them as one group at different maturity levels, segmented by experience rather than by asset category.
Serving users across more than 170 markets
MEXC now covers more than 170 markets, and Lin said its user-layering approach has both visible and invisible components.
The less visible side depends on stronger analytics tools and big data systems. Once a user enters the platform, she said, MEXC can infer which channel brought that user in, whether the profile looks more like a traditional finance participant or a higher-risk trader, and whether the person behaves more like a trader or a depositor. In the past, user profiling across the industry was crude, often split into just three buckets: beginners, institutions, and VIPs. In reality, many VIP users behaved much like institutions, but the industry lacked the data and analytical tools to segment them properly.
Now the profiles are refined into more than a dozen categories. Lin gave one example: for a conservative user who only trades BTC and ETH and cannot tolerate deep drawdowns, the AI strategies shown would lean toward major-asset trend strategies, trendline signals, or reversal signals with smaller drawdowns and backtesting support.
The part users can see is a new VVIP system. Traditional VIP programs usually rank users from level 1 to level 9 based on deposits and trading volume. MEXC’s VVIP structure adds M Score, which is not based only on assets. Lin said it converts all user behavior on the platform into a score, functioning more like a loyalty card. Daily logins, campaign participation, and trying new products can all add to that score.
Users with stronger loyalty, she said, can receive more early access to new products, better entry and limits on high-yield wealth products, and higher cashback rates on the MEXC card. The VVIP system is linked to the broader user profiling system through M Score, and Lin described the rules as relatively transparent and more advanced than many market alternatives.
How localization feedback changed internal tools
When asked for a concrete example of localization feedback driving product iteration, Lin did not point to a front-end feature. Instead, she highlighted a large effort over the past three months to rebuild internal operations tools.
Previously, she said, operations staff often had to configure campaigns based on personal judgment, and that made outcomes heavily dependent on experience and generation. A veteran operator with 10 years in the field, a new recruit, someone born in the 1980s, and someone born in the 2000s could all read users very differently. Across more than 170 countries and regions, with different age groups and risk preferences, precise manual operations were close to impossible.
MEXC used AI and big data to refine user profiles and then built those insights into atomic internal tools through data engineering. The result, she said, is that operations staff can now launch and configure campaigns much faster.
For different countries and partners, a campaign tailored to current events and local community preferences can now usually be configured in under an hour. In the old process, writing copy, designing visuals, and localizing into multiple languages could take two or three days. Lin gave the example of a sudden market surge the day before. In the past, it would have taken at least two or three days to move from idea to launch. Now the full process can be completed within 30 minutes, allowing users to engage with the market while the move is still fresh and while incentives are still timely.
The hidden product work under the surface
Lin said a lot of what supports retention and conversion is not immediately visible to users.
One major area is technical optimization. MEXC is in the middle of a large system upgrade, she said. Once completed, users should feel two direct changes: faster app opening and faster page switching. Sensitive users may notice the speed gain clearly. Most people will simply feel that the app is smoother without being able to explain why. For the infrastructure team, she said, the effect is much bigger than it looks from the surface.
The new base layer is also meant to handle larger trading throughput, reduce the risk of outages during extreme volatility, and support future work in lending and options. Lin described that as a form of psychological safety for users.
Another hidden layer sits in account structure and trade requirements. She used AI strategies as an example. Users may only see that some strategies still perform reasonably well even in a deep bear market. What they do not see, she said, is that a product team of four to five people spent three months studying the styles and strategies of historical investing masters one by one, then adapting them to the volatility and behavior of crypto assets.
What MEXC is measuring from its AI tools
Lin divided MEXC’s AI products into several groups.
The first group includes products users can clearly perceive and approach with a defined goal, such as MEXC AI and AI strategies. MEXC AI is more information-oriented, she said, acting like an information manager that gathers global news and the technical indicators users need to watch. The team looks first at whether people are actually using it. In Lin’s words, users vote with their feet: if the product works, they come back every day to ask questions and check updates. Once it becomes part of a user’s daily information flow, the platform has more chances to surface hot tokens and current trading opportunities and, from there, convert activity into trading.
AI strategy tools sit closer to execution. Lin said users who reach this feature are usually near the point of building a trading system or placing an order. The team mainly tracks two indicators: repurchase or repeat-use rate, and how many users actually build strategies and run backtests. Those numbers also help MEXC understand who on the platform is relatively sophisticated and who opens the feature, finds it difficult, and leaves. In the latter case, the team still has to decide whether the problem lies in the product or in a mismatch between the user and the intended audience.
A second group includes efficiency tools such as AI News, AI Search, and a command-line interface that is about to launch. Lin said these tools are less about immediate trading conversion and more about preparing the interaction model for the next generation of products.
She used AI Search as the clearest example. The function is designed to solve end-to-end navigation. In the old flow, a user entering from the homepage and looking for a wealth product had to find the correct menu first, then browse a long list. A user looking for a less common asset might have to scroll for a while and could still miss it if several assets had similar names. On a phone, where screen space is limited, many people stop scrolling after a few pages. Useful products get buried.
Now, if a user types BTC into the search bar, the system can return the three most relevant news items, the current price chart, technical indicators, ongoing campaigns, and whether BTC has wealth-management or yield-related offerings on the platform. Lin said that amount of content cannot be assembled by hard-coded manual logic alone. AI is now involved, at least to some extent, in component orchestration, interface calls, and content selection. Search one term, she said, and the user can see almost everything on the platform tied to it.
From assistant to copilot
Lin broke AI development in trading into three stages.
The first is the assistant stage: the user asks one question and the system gives one answer.
She said the market is now entering the second stage, the copilot phase. Under prior user authorization, AI can understand what the user is doing and stay present through the process. Lin gave an example: if the system learns that a user tends to make impulsive decisions during sharp market drops, then the next time that situation appears, it could intervene and suggest something different. Instead of buying a high-leverage product at that moment, the user might be prompted to buy spot or consider a high-yield deposit product for a more certain return.
Further out, she expects a more autonomous form of operational intelligence that changes how trading is carried out. Today, users still have to open the app and navigate to the right page to perform a function. In the future, she said, there may be one unified gateway, such as embedding MEXC’s assistant directly into a phone operating system and linking it with Siri. A user could say: “BTC looks strong today. Can you help me make 1,000?” Siri would then call MEXC’s command-line interface, connect to back-end services including backtesting, assess the user’s current assets, and determine whether that target is achievable that day. It could execute directly, or if uncertainty remains, present three choices for the user to pick from.
Lin compared that scenario to a highly intelligent world closer to The Matrix. The biggest barrier, she said, is computing power. Application-layer companies can make interactions cheaper, more efficient, and more accurate, but they cannot fix hardware and infrastructure limits on their own. Once those constraints ease, she expects product experience to change dramatically, potentially within one to two years.
What MEXC wants to build over the next year
Looking ahead, Lin said AI has already penetrated every stage of MEXC’s product and R&D work. But when it comes to user-facing products, the company still wants to stick to a first principle: keep the user at the center.
MEXC does not need users to think it is the platform using the most AI, she said. The goal is to spread the company’s understanding of AI and of users quietly through every small function and every piece of information. The ideal result is simple: users may not even realize how many AI features they have used, but they feel the platform understands them better.
To get there, she said, MEXC still has to finish a lot of groundwork. First, AI has to understand assets better, which means continuous learning from markets. Second, it has to understand users better, which depends on user profiling and big data. AI is good at analysis, Lin said, but it cannot by itself complete the groundwork of data storage, cleanup, and standardization. Human effort still has to build the base before AI can do its best work.
She also said product direction must move with the market. During the recent softer market period, trend-following products were less appealing, which was one reason MEXC pushed AI strategies. Those strategies include both trend-oriented models and models built to capture spreads during bottom-range consolidation. Because the company had already done technical and data preparation, Lin said, rolling out those products was relatively smooth.
The next three years: preparing for users coming from traditional finance
Asked what fundamental changes she expects in exchange products over the next three years as crypto and traditional finance continue to merge, Lin said MEXC’s central task is to serve future users coming from traditional finance into crypto.
Customer service, BD, partners, and product strategy teams are all spending more time trying to understand what those users want to trade, which tools they expect, and whether MEXC’s technical infrastructure can reach a level of speed and product breadth comparable to traditional finance. Lin called that the most important strategy for this year and next year.
Under that strategy, operations teams need to pay closer attention to traditional finance assets, while BD teams need to win traditional finance users and channels. For product teams, one major issue is filling gaps in tools that traditional finance users already take for granted but crypto still often lacks.
In Lin’s view, crypto does not have a shortage of major product categories. It is short on detail. She gave iceberg orders as an example, calling them a popular order type in traditional markets that some crypto exchanges support and others do not. Those are the smaller but important features MEXC wants to add quickly, so that when traditional finance users arrive in greater numbers, the platform will be able to meet their expectations.