Binance has released AI Pro, and I tried it out with Hynix. I didn't ask it to predict the rise and fall, nor did I ask 'can I buy now'. I gave it a specific requirement: can the premium between Hynix ADR and Korean stocks be made into an automated trading strategy? Both of these have contracts on Binance, with SKHYUSDT corresponding to ADR and SKHYNIXUSDT corresponding to the Korean listed Hynix. My idea is simple: after unifying the stock exchange ratio, exchange rate, and contract unit, if the ADR premium reaches 45%, then go short of ADR and buy more Korean stocks; The premium returns to 35%, and both sides are tied together. Let's not discuss whether these two parameters are reasonable for now, nor do we recommend everyone to follow them. I just want to see where the current exchange AI can take this matter to. As we talked, it really understood what I was saying. It can list the contract unit, conversion ratio, trading depth, slippage point, and funding fee. Even if one side makes a deal and the other side doesn't, it will remind you what to do. As an assistant for discussing strategies, I think it is useful. Especially when you only have a rough idea in your mind, chatting with it for a few rounds can fill in many unclear areas. But what I really want is still ahead. I asked: Can you directly help me make a Bot, run it in Binance, and trigger the condition to trade? It told me that currently it is not possible to directly deploy a continuously running Bot here, nor can it automatically start the live drive for me. Then continue asking me if you want to help me write technical specifications and design a backtesting framework. Seeing this, my feeling is that we have talked about the plan almost, and now I still have to find someone to work with myself Of course, it needs to be made clear here. The entrance in my screenshot shows' Binance AI ', and the official introduction to the new version of AI Pro includes strategy building, simulation, and real deployment. So this time, I can't be sure if the entrance hasn't been opened correctly, the functions haven't been fully activated, or if this two legged strategy is not supported. But for an ordinary user who uses this and is unsure if they can do it, it actually indicates that there are still areas where the product needs to be polished. What I hope Binance can add the most is to connect this step. For example, when I finish talking about the Hynix requirement, it directly tells me: Can these two contracts be used? Does the dual leg strategy support it? Is the data sufficient. If possible, generate a strategy chart that displays opening, closing, positioning, and exception handling. If there are any missing parameters, continue asking me. Then simulate running first. When was it triggered, based on whether it was possible to close the deal at that time, how much funds were left after deducting the fees on both sides, and what would happen if there was no transaction at all? Please write it down and show me. These things are more useful than giving me a lengthy strategy explanation. People involved in trading should understand that sometimes you don't lack an idea, what you lack is the time and energy to implement it. A seemingly simple price difference strategy, if you really want to do it yourself, you have to take market trends, check units, write programs, find servers, and handle errors. Halfway through, the opportunity may have passed. If these tasks can be completed through dialogue in Binance, I am willing to use them. And I have expectations for Binance to do this. Market trends, accounts, and orders are already on the platform, and if they can be well managed, users can indeed save a lot of trouble. But perhaps the most difficult thing here is precisely these things behind. Explain why opening a warehouse is easy. What should I do if I encounter a situation where I get stuck while making a deal? The strategy has been paused, is the position still available? The market data is broken, should we continue placing orders or stop? Real gold and silver are running, and these details need to be explained. I don't expect AI to find me a profitable strategy either. A 45% premium can become 55%, and simulating profits can also result in actual losses. Whether the strategy has advantages or not still needs to be researched by oneself. What I want to hand over to it are the rules that I have already thought through, but I am too lazy to stare at them repeatedly. This experience helped me sort out the problems more clearly, but the Hynix Bot has not yet been developed. If I can really connect it later, I will continue to try the same example. After all, for me, to judge whether a product is good or not, there is only one thing to consider in the end: After I finished speaking about the requirements, was it an additional manual or a tool that could help me with the work.