Why does the AI trading assistant need to give users an identity card first? View the product demonstration on the NeoTrade Web webpage

After AI starts executing trades for users, industry discussions often focus on model capabilities and the level of automation. However, when users open an AI trading product, a more immediate question is: who is this AI, how does it make decisions, and how do I know what it is doing? The web interface is often where users first encounter the product, and what is displayed on the first screen determines the expectations users carry forward.

Launched on November 3, NeoTrade Web provides a sample worth discussing. This is the first time NeoTrade has a web version; previously, it only had a desktop client. Its process is clearly arranged: first, users get to know themselves, then the AI, and finally, they enter trading. The most noteworthy step is that before trading begins, the AI is given an on-chain identity card.

1. After lowering the threshold, the web needs to clarify itself first

From client to web, the most direct change is the entry point: no need to download and install, just open a browser to use it; logging in is no longer contingent on having a wallet, and users without a wallet can also enter.

With the threshold lowered, the incoming users will be more diverse, many of whom may not be familiar with on-chain products. This places another requirement on the web: it must clarify itself within the first few minutes. A shorter entry does not necessarily mean better retention; what truly determines whether users stay is still what they see after they enter.

2. Let users describe themselves first

After entering NeoTrade Web, users first complete a trading style assessment, resulting in a six-dimensional trading profile.

The significance of this opening lies not in how accurate the assessment is. Trading styles are difficult to fully define with just a few questions, but the assessment provides users with something they can confirm or refute. For an AI that will act on behalf of the user, having users describe themselves first is closer to the order of trust-building than simply showcasing what the AI can do. Of course, whether the profile reflects true preferences and to what extent the AI's execution aligns with the profile will ultimately need to be validated by actual trading records.

3. The identity card answers the question of who this is

After the assessment, users generate an ERC-8004 identity card for their AI trading agent, creating a verifiable on-chain identity for the trading profile. This is the first time NeoTrade has put AI identities on-chain. Meanwhile, all agents are uniformly displayed with dynamic 8004 identity cards, which are linked to the agent's automated trading style, making it easy to see what style it represents.

There are two layers of value here. First is recognizability: when each agent appears in the same format, abstract trading preferences become identities that can be compared side by side. Second is verifiability: allowing the AI to trade essentially means delegating part of the judgment. If the AI's style of operation only exists in backend parameters, users can only see the results; putting identity on-chain turns this delegation relationship into a publicly verifiable object.

However, it is important to distinguish that verifiable identity and trustworthy trading performance are two separate issues. The identity card can answer which agent it is and what style it is associated with; as for whether it executes according to the style and how it performs, that requires trading data and corresponding verification mechanisms to answer. Just having the identity on-chain does not prove an agent's trading capability.

4. The deeper the automation, the more visible the intent needs to be

Once the identity is established, the agent will continuously and automatically trade in prediction markets according to the user's assessed style, while pushing NPT point tasks and proposed transaction cards to the user.

Continuous automated trading means users cannot monitor every transaction. The value of the proposed transaction card lies in presenting what the AI intends to do in a readable format. The higher the level of automation, the more important the visibility of intent becomes.

It is also worth noting that NeoTrade Web currently does not support dialogue. For AI products, this may seem like a missing piece, but it makes sense in terms of the order of presentation: when identity and style have not yet been established, open dialogue can easily lead users to imagine the AI as an all-knowing advisor. First, clarify the boundaries with assessments, identity cards, and proposed transaction cards, and then consider where to place dialogue; this order is more prudent.

5. Separate incentives from usage

Another easily overlooked design is that NPT points and AI quotas are recorded in two separate accounts. Points accumulate with tasks, while quotas are used up with each use; when insufficient, they are supplemented by non-transferable SBTs.

When incentives and usage are mixed into one number, users only see a vague balance; once separated, it becomes clear what actions were taken, what was gained, and how much the AI can still use. The proof for supplementing quotas cannot be transferred, making the quota more akin to a usage right rather than a transferable asset. For AI trading products, clear accounts are part of building trust.

Conclusion

Returning to the question in the title. The AI trading assistant first gives users an identity card, with the focus not on the card itself, but on the order it arranges for users to enter the product: first confirm what style they are, then confirm who the AI is, and only then see what the AI has done. The display space on the web is limited; it is easy to fill it with functions, but arranging the order correctly is more challenging.

However, if the identity card is only used for display, its value remains at the interface level; only by accumulating continuous, verifiable behavioral records around the same identity can it potentially transform from a card into a basis for judging the agent. Whether such designs can reach this point and form a habit of continuous use still requires the product's operation and real user behavior to provide answers.

Risk Warning: Trading in prediction markets involves risks, and AI trading may result in losses. This article does not constitute investment advice.