What is zScore

zScore is a decentralized reputation system that quantifies wallet reputation through AI-powered behavioral analysis.

zScore is a decentralized reputation system that quantifies user reputation in the onchain economy. It uses a behaviour centric neural network that computes reputation scores based on:

• Onchain interactions with DeFi protocols

• Historical transaction patterns

• Behavioral analysis through AI neural networks

• Verified offchain credentials (via zkTLS)

Problems with Onchain Reputation

We have seen several attempts to tackle identity and reputation problems across Web3. From decentralized identity solutions like Civic and Bloom, to more recent attempts like Cred and RociFi, every solution has drawbacks whether it is poor uptake, highly centralized and permissioned systems, small datasets, or usage limitations. They haven’t fully bought into building on decentralized systems, or thinking really broadly about the reputation space.

Some solutions rely primarily on proprietary behavioral analysis data sets and algorithms that are expensive to access. This becomes a problem for teams and protocols to implement solutions that may or may not apply to their needs, and it’s not possible for end users to trust that the underlying data assesses their reputation fairly.

Others focus heavily on identity verification. This is a tried-and-true way of confirming someone’s identity is highly intrusive and expensive. Additionally, there are many pieces of valid ID proof that aren’t accepted by such systems, requiring manual review or simply being rejected outright even though they are accepted locally.

There are also reputation systems focused on specific niches. These can be useful, but they are not adaptable to new use cases. E.g., a reputation scoring system focused solely on one blockchain might not contain relevant data for another blockchain, or one focused on DEX activity might not take into consideration behavioral differences on a lending platform.

Regarding traditional credit scores (e.g. TransUnion, Equifax, CIBIL), many news articles, papers, and other reports outline the many flaws they contain and the trouble they have caused individuals around the world. Traditional credit scores can be useful signals, but they have to be matched up with a wallet in order to be applied to Web3.

zScore as the Solution to these Problems

zScore is a public good that computes your onchain reputation score based on your past borrowing and repayment history, now live on Base mainnet. This is thanks to EigenLayer’s Actively Validated Service (AVS) platform, being operated by a decentralized set of operators who are running the zScore infrastructure.

By leveraging AI and decentralized technology, zScore provides a transparent, unbiased, and secure way to assess reputation in the blockchain ecosystem. Generating and updating zScores is done at the wallet level, and does not require identity verification or KYC processes. This is a verifiable, AI-driven neural network score that combines both onchain and offchain reputation to help people unlock access to a better blockchain experience across a variety of protocols and situations, and allows builders to easily suss out real, high-quality users from the suspect ones and bots.

zScore v1 has been trained on the onchain history of 1,250,000 wallets’ usage of the Aave protocol since 2020. This gives us an incredible dataset to start with, with many more to come in the next version of zScore including other lending protocols, DEX activity, perpetuals trading, and even staking transactions.

zScore also takes both offchain reputation and onchain behavior into account, running meaningful onchain transactions and interactions through behavior-centric neural networks to arrive at each wallet’s zScore. In this way, we can identify positive versus value extractive interactions, where the latter are flagged and penalized. We can also bundle offchain creditworthiness thanks to Reclaim Protocol’s implementation of zkTLS, binding real-world data to one's wallet in a secure, privacy-preserving and opt-in manner.

zScore can be considered as a method to derive reputation without the need to use social authentication or full KYC.

zScore is generated in a verifiable manner through the zScore AVS, running on EigenLayer, built with Othentic Stack. This makes it possible to not only run the AI model in a verifiable and repeatable manner, it makes the operation of zScore truly decentralized thanks to the group of AVS node operators that have already signed on to run it, and it’s back with the crypto-economic security of EigenLayer’s billions of dollars in staked Ether.

Additionally, we’re leveraging LayerZero’s OApp standard to make zScore easily readable and updateable across dozens of blockchains.

As a freely accessible public good, zScore can be and is now a key part of any project’s tech stack to help differentiate itself from established players whether they have been building for 6 weeks or 6 years.