0xScope: AI-Driven Insights and Innovations in Web3 Data

What is 0xScope?

0xScope is a protocol to become the AI data layer for Web3 applications, offering a suite of SaaS on-chain data solutions. The 0xScope ecosystem primarily consists of three products:

• Scopescan, a blockchain data analytics platform based on knowledge graphs;
• Scopechat, an AI assistant platform based on large language models;
• ScopeAPI, aimed at enterprises and developers, can differentiate between real users and bot accounts, understand on-chain behavior, and aggregate addresses, thereby providing a certain degree of Sybil attack monitoring.

Its target audience is very broad: Web3 venture capital funds, protocols and Dapps within various ecosystems, including Layer 1 and Layer 2, venture capital firms that need to make better investment decisions and monitor competitors, traders/researchers who need to investigate capital flows, market makers, developer communities, and general investors.

Team Background

Jackson Li is the founder of 0xScope, with a background in data analysis and product management; Co-founder Pedro Torres serves as CEO, with a rich background in on-chain transactions. Another co-founder, 0xWill, has a background as a researcher and product manager. Other key members include co-founder Colin Yu as CTO, and RAGINI Raffaele as CMO.

Investor Analysis

Starting in July 2022, Watchers (later renamed Scopescan) began a trial operation with a whitelist application system and has since been iterating on the product based on user feedback. According to public data compiled by rootdata, 0xScope has completed three fundraising rounds, raising 8 million US dollars from 14 investment institutions or individuals. In September 2022, deep-tech news outlet TechFlow reported that 0xScope completed a 3 million US dollars seed funding round; on January 16, 2024, it completed a 5 million US dollars pre-A funding round led by HashKey Capital.

Source: Table data from rootdata

The Main Products and Features of 0xScope

Scopescan

Scopescan utilizes the visual effects of “Knowledge Graphs,” a technology based on graph theory for knowledge representation and reasoning. This aids users in intuitively examining the results of address clustering. Here, we will start with the address clustering function to introduce the main features of Scopescan.

Address Clustering: Address clustering refers to aggregating and marking addresses that may come from the same entity on the blockchain (since a single address is insufficient to depict an accurate picture, Web3 users often control multiple addresses). This allows us to understand better the entity’s blockchain behavior, the proportion of tokens held, etc. These factors may influence governance rights within a community, proposal rights, and our understanding of smart money, VCs, and whales.

The address clustering feature of Scopescan involves two concepts: one is the tracking of addresses, and the other is the definition of entities. First, as of the time of this article, 0xScope has tracked a total of 176,719,155 blockchain addresses; second, Scopescan uses an original weighted clustering algorithm that can detect the association between different EOA addresses and define entities representing more real blockchain subjects, with the confidence interval adjustable between 70%-90% accuracy. The principle of this feature is to match the blockchain behavior between addresses with the relationships in the label library. Below is the product interface of the Scopescan address clustering feature.

Source: Scopescan Address Aggregation

• Smart Money: Aggregates “smart money” address details, supporting users in viewing and analyzing alerts from various dimensions, and even filtering out certain addresses to create a separate entity for tracking and alerts.
• ScopeVC: The data layer provided by Scopescan can also be used for B2B services, creating enterprise-level customized products. An example is ScopeVC, a B2B tool for venture capitalists for pre-investment research and post-investment management.

Scopechat

Scopechat is an AI assistant developed by a large language model, providing real-time on-chain data and analysis. Currently, it covers blockchains, including Ethereum, BNB Chain, Arbitrum, Polygon, Base, Optimism, Avalanche, and Mantle. After its Beta phase, Scopechat was opened to the public on January 16, 2024, and supports mobile devices. Currently, registered users can ask 25 questions per week, while unregistered users can ask 3 questions daily.

According to the user interface introduction, the main functions of Scopechat are as follows:

• Strategy Overview: Trend trading is a strategy where traders make decisions based on the current market trend, expecting the trend to continue to profit.
• Target Audience: Short-term traders can focus on token price trends, entering and exiting the market quickly with small positions to profit from minor fluctuations. Long-term traders can analyze narrative heat and technical aspects to identify longer-term trends.
• Risks: Trends may suddenly reverse, so it is crucial to manage risk by taking timely profit-taking and stop-loss measures. Also, market noise from news and rumors can disrupt trend trading decisions.

Source: Scopechat User Interface

What Problem Does 0xScope Solve?

Democratization and Decentralization

Address aggregation aids in our understanding of the degree of centralization in the on-chain world, contributing to the realization of the vision for Web3 democratization and decentralization. A classic example is the governance token $UNI of the Uniswap protocol. In theory, $UNI holders can propose the future development of the Uniswap ecosystem based on the community’s ideals and visions. However, in the snapshot below from September 29, 2022, it can be seen that although a16z’s wallet holds 1.5% of $UNI, the aggregated addresses associated with this wallet account for a total of 4.15% of this token, and the Uniswap decision-making threshold (Uniswap quorum) is 4%, meaning that a16z could independently propose initiatives.

Source: snapshot of bubblemaps.io data

0xScope also shares the vision for the democratization and decentralization of Web3 data. A corresponding example is the collaboration between LooksRare and Layer3. In the snapshot from December 15, 2022, we can see an increase of 3,280 new addresses, with a corresponding entity count of 2,683, which is approximately 25 times higher than the natural user growth rate. This indicates that the collaborative event was very effective.

Source: Scopescan

The issue is that retail investors want to know whether these new addresses are operated by professional bonus-hunting teams managing numerous wallets to participate in events, or if they stem from genuine users. Data aggregated by Scopescan shows that among the 3,280 new addresses interacting with LooksRare, there are 2,685 entities. The entity-to-address ratio is about 82%, indicating that most represent real users.

The Value of High-Quality Data Layers

0xScope’s approach involves leveraging aggregated tags and establishing a comprehensive standard for collecting, cleaning, and managing related on-chain and off-chain data through a Web3 AI data layer. This requires the team to gradually build an open-source data layer, enabling people to upload, verify, process, and download Web2 and Web3 data.

These visions are interlinked: given the anonymity of on-chain data, analyzing it is often very complex and noisy. Therefore, 0xScope aims to filter the noisy data into high-quality data; a major challenge in integrating AI into Web3 is the lack of high-quality data. They can use high-quality data sets to train their AI assistants and develop this service into a B2B product.

Source: 0xScope Data Layer Architecture

AI Assistants and Mass Adoption

0xScope utilizes the data layer employed by Scopescan to train Scopechat, a large language model (LLM) centered around Web3 use cases. It enhances the model’s accuracy through continuous feedback and fine-tuning. Developers can also directly train Web3 AI assistants based on this model. The primary challenge of blockchain technology is its high entry barrier and low adoption rate. Scopechat aims to boost the efficiency of processing Web3 transactions, data, and contract interactions through large language models and AI assistants, promoting mass adoption. Essentially, this approach seeks to create “GPTs for the crypto sphere.”

Challenges of Integrating Large Language Models

Training large language models (LLMs) continues to face significant challenges in combating hallucinations and improving accuracy. Colin Yu, one of the co-founders, has made two key analyses on Twitter:

1. LLMs struggle to understand the data structures of Web3, making it impossible to guarantee 100% accuracy in AI-generated SQL outputs. The most viable approach is fine-tuning, but this binds the database structure to the model, leading to increased update costs, and decreased accuracy of outputs when the model has fewer parameters.

2. Beyond fine-tuning, strategies like Embedding+searching, and the LangChain project’s agent+tools approach are promising. The Embedding technique, highlighted by last year’s popular Alexander Project, proves exceptionally versatile, covering functions such as search, clustering, recommendation, and classification more effectively. For instance, searching for a paper by keywords alone might not yield results synonymous with those keywords; Embedding easily overcomes this challenge.

The LangChain project utilizes different Agents (LLM + prompt) for specialized tasks. These Agents can also call upon various Tools (similar to Functions) tailored to unique uses and scenarios.

Potential Airdrops and Conclusion

The product Scopechat, having been launched just over two months ago, is still in the process of continuous iteration. Therefore, besides acquiring Scopepoints through basic referral codes, linking social media accounts, and joining communities, providing effective feedback for Scopechat also earns corresponding Scopepoints. According to the official documentation, it is confirmed that these Scopepoints can be used to purchase products and services from 0xScope or unlock advanced features of Scopechat.

In the FAQ section of the official document, regarding the possibility of issuing a cryptocurrency, the official stance prioritizes the community and the product itself but hints that “everything else will follow” if the product is on the right track, not ruling out the possibility of a token launch. Furthermore, Scopepoints may also be related to potential airdrops.

In summary, 0xScope has also developed an AI assistant besides services for on-chain data analysis, capitalizing on the trend of large language models. This is highly competitive in the fast-paced bull market scenario.