XT Chief Strategy Officer and Head of XT Labs Aaron.J discusses what to delegate to AI agents and which decisions traders need to keep.

Researching a trade, watching the market, and placing an order require different levels of access. As AI tools begin to make judgments and act on a trader’s behalf, users need to decide where that authority ends.

During XT’s Chinese-language AMA, “The Next Step in AI Trading: Exploring Agent Applications Through NeoSoul (DNA),” XT Chief Strategy Officer and Head of XT Labs Aaron.J put the distinction this way: “The difference isn’t whether it’s automated. It’s who makes the decisions, who makes the judgment.”

That distinction runs through his approach to using AI. Traders can delegate individual tasks gradually while retaining control over position size, leverage, and how much they are willing to lose.

Who decides when an order is automated?

Aaron.J compared grid trading with copy trading. A grid buys and sells according to rules set by the user, with execution handled automatically. In copy trading, the lead trader makes the main trading decisions and others follow those trades. An agent authorized to analyze information and act independently takes on some of the decisions that would otherwise belong to the trader.

All three may place orders automatically, but each asks the user to give up a different degree of control. Following preset rules is different from revising a trading decision in response to new information. Permission to collect information does not, by itself, include permission to use funds.

Aaron.J favors granting that authority gradually. He sees uses for AI in gathering information, monitoring markets, and reviewing past trades. He is more cautious about letting it decide what to trade. In his view, data and market conditions can still affect its judgment, making it too early to hand over full trading authority at once.

Start with fewer mistakes and less routine work

Asked what he most wanted AI to help with, Aaron.J described a familiar trading problem: a trader sets an exit condition, then hesitates when the price reaches it, hoping for a rebound. He wants AI to help people follow the rules they have already set, reducing the mistakes that come from changing their minds in the moment.

Research is another place to save time. Collecting information, tracking price changes, and reviewing trades take sustained effort. Aaron.J would like tools to handle the repetitive parts, leaving traders more time for decisions that need their attention.

He compared this division of work to a trading firm, where researchers, strategists, risk managers, and traders each have a role. The same approach can help users choose an AI tool: identify the task they need help with, then look for that capability. A tool that organizes information well can help with research without needing permission to place trades.

He would start with information gathering, add help with monitoring risk, and then consider granting permission to execute trades. Even during testing, he favors limiting the amount of money involved. Important settings such as position size and leverage remain under the user’s control.

Define when permission ends

Aaron.J identified two requirements for granting access: a financial limit and a clear scope. Users need to establish how much money a tool can access and which actions it can take. He also discussed when a system should stop during periods of sharp market volatility. Along with the ability to act, a system needs clear conditions for ending that authority.

Toward the end of his remarks, he repeatedly urged listeners to establish how much they could afford to lose and what activity should stop when that limit is reached. He wants users to settle those questions before using AI, rather than waiting until a trade goes wrong.

In product terms, a loss limit needs to be backed by actual permission settings and enforcement. Writing an instruction to a model does not automatically create a reliable risk control. Users need to know how a stop instruction takes effect, which actions it restricts, and how they can revoke access.

Aaron.J is optimistic about what agents may eventually be able to do, but his starting point is cautious. Information gathering and research can be delegated first; important decisions involving money stay with the trader. Whether to grant more authority depends on exactly what the next step hands over. A user should understand how to stop a trade as clearly as how to let a system place one.

---

This article is based on Aaron.J’s remarks during XT’s Chinese-language AMA, “The Next Step in AI Trading: Exploring Agent Applications Through NeoSoul (DNA).” Quotes are translated from Chinese. This article does not constitute investment advice.

Listen to the full replay.

About XT Exchange

Founded in 2018, XT Exchange is a global digital asset trading platform serving more than 12 million registered users across over 200 countries and regions. Its services span spot, margin and futures trading, TradFi market exposure, a real-world asset marketplace, and everyday payments through XT Pay. Guided by Xplore Crypto Trade with Trust, XT Exchange enters its ninth year with Build the NeXT, reflecting its focus on trust, broader market access, and practical uses for digital assets.

Join the XT Exchange Community: X (Twitter) | Telegram | Facebook | Instagram | LinkedIn | Medium | YouTube

Disclaimer: XT Exchange reserves the right, at its sole discretion, to modify, amend, or cancel this announcement at any time for any reason without prior notice.