Your AI Needs Your Data: Building a Personal Data Layer with DataDance Wallet

 

AI is already helping people make decisions.

Many people now turn to AI tools and AI assistants when they want to buy something or plan an activity. Instead of browsing dozens of pages, they ask AI to generate ready-to-buy product lists or full travel plans based on their needs.

In one Reddit discussion, a user shared that they had already used ChatGPT to decide purchases such as a trampoline, socks, a toaster oven, and even their next coffee maker. In another thread, users discussed how to craft prompts that help ChatGPT recommend specific products, such as shampoos or grooming items.

These examples show how AI is gradually becoming part of everyday shopping decisions.

However, some discussions also show that these AI-generated suggestions are not always reliable. In the trampoline example, one user replied that when they asked ChatGPT for product recommendations, it returned items that did not even exist on Amazon. In the discussion about prompting ChatGPT for grooming products, several users noted that the results often depended heavily on how the prompt was written. A small change in wording could lead to completely different recommendations.

This suggests that if we want AI to act as a real personal assistant or AI agent, current interactions are still not enough. Users often need repeated prompts and adjustments before the results begin to match their real needs.

Why does this happen?

AI Still Doesn’t Really Know You

The main reason is that most AI systems today do not actually know the user.

Whether it is large language models or emerging AI agents, the data used to train them is mostly generalized public data rather than individual personal data about a specific user.

For AI systems, collecting personal behavioral data directly from individuals is also difficult. Current data markets typically sell aggregated datasets organized by category or platform, rather than continuous behavioral histories tied to a specific person.

This means AI models can access large amounts of public information, but rarely have access to the long-term behavioral context of a particular user.

Another reason is that users themselves do not have a single place where their personal data is organized.

Today, most digital activity records are stored separately across different platforms. Shopping history may exist on Amazon. Travel bookings are stored in services such as Airbnb or Booking. Event participation records may appear on platforms like Luma. Each platform holds only a fragment of a user’s overall activity history.

Because these records remain locked inside individual services, users cannot easily gather them into a unified dataset. As a result, even if someone wants to use an AI agent as a personal assistant, they cannot simply provide their complete digital history to the system.

Without access to continuous personal context, AI systems can only rely on prompts and generalized knowledge. This is why even powerful models still struggle to generate highly personalized suggestions.

DataDance Wallet as a Personal Data Layer

This gap is also beginning to attract attention across the AI industry.

AI researcher Andrej Karpathy recently pointed out that the next stage of AI development may depend less on model improvements alone and more on access to personal context. In his view, AI systems will eventually need structured personal data rather than relying only on prompts and fragmented tools.

This idea has resonated with many builders and researchers. Some have started describing this emerging layer as personal context infrastructure — a system where users maintain a unified data layer that AI systems can access with permission.

Such a layer could include shopping history, travel activity, professional events, preferences, and other long-term behavioral records that together form a user’s real digital context.

This is exactly the type of infrastructure that DataDance Wallet is beginning to explore.

DataDance Wallet is designed as a decentralized data wallet where users can upload selected digital activity records and earn rewards, similar to how traditional receipt reward apps allow users to upload purchase receipts. Instead of scanning only paper receipts, DataDance Wallet focuses on digital activity records that already exist across different platforms.

When users choose to upload these records, the system processes them into structured data while applying privacy-preserving technologies. Sensitive information can be protected through mechanisms such as secure computation and user authorization. This allows the data to be verified and used without exposing raw personal information.

At the same time, users receive rewards for contributing these records, turning everyday digital activity into a new type of user-owned data asset.

Under this model, DataDance Wallet is not only a reward system for digital receipts or activity records. It also begins to function as a personal data layer where users can gradually bring together different pieces of their digital activity history.

At the moment, this layer is being built step by step through real activity records.

Users can already upload Amazon order history to contribute shopping data. Support has recently expanded to platforms such as Airbnb and Booking, allowing travel-related records to become part of the same dataset. Event participation records from Luma can also be uploaded, adding another dimension of professional and community activity.

Together, these records begin to form a permissioned dataset of real user behavior across shopping, travel, and community participation.

This kind of structure creates something that traditional AI systems currently lack: a user-authorized source of long-term personal context.

Rather than relying only on prompts or generalized datasets, AI systems could eventually access structured activity records that reflect how a person actually shops, travels, and participates in communities online.

Toward a Personal Data Layer for AI

Looking ahead, while shopping activity, travel bookings, and event participation already reflect a large part of everyday digital life, they are only the beginning.

To support a more complete personal data layer, DataDance Wallet plans to gradually expand the range of data sources and platforms that users can contribute from. At the same time, the project is also exploring how this user-authorized data layer could connect smoothly with future AI agents, allowing individuals to provide structured personal context directly to the systems that assist them.

If AI is to become truly personal, the data behind it will also need to become personal.

If you would like to learn more about DataDance Wallet and upcoming platform support, you can follow the official channels below:

APP | Official Website | X | Facebook | Linkedin