TLDR;
-
Recommendation systems enhanced by AI and LLMs are becoming increasingly sophisticated, but currently lack data sharing between platforms. Blockchain networks could provide the infrastructure to improve these systems by making user activity openly readable across an entire network.
-
Lucidity is building a Web3 recommendation system that operates across three dimensions (assets, protocols, and wallets) to provide personalized suggestions for DeFi strategies, governance opportunities, and community interactions.
-
The integration of AI in blockchain systems requires verifiable frameworks (like opML) to maintain decentralization principles and prevent exploitation through opaque AI systems.
Recommendation Models
Recommendation systems are integral to many modern sector behemoths of the Internet. Evolving from personal email filtering to matrix factorization enhanced by machine learning, today they are increasingly sophisticated. Modern day recommendation systems more closely resemble AI systems due to the benefits LLMs provide in improving adaptability and contextual understanding. Deep learning models also address challenges with data sparsity and noise. Together, these AI-based improvements have made recommender systems more personalized, efficient and interactive, therefore making the applications in which they’re embedded more engaging for users.
E-commerce, media and advertising are just a handful of the sectors that have benefited from the development of these models. Recommendation models can be used to improve user experiences (UX) across the internet, should data be shared and the user profiles generated be comprehensive enough. Traditionally, there has been a lack of information flow and data to inform recommendations between platforms.
Blockchain networks, which act as an open source ledger facilitating user activity, may well provide the infrastructure to amplify the power of recommendation models. All user activity is openly readable, and recommendation models can be designed for assets, applications, users and interfaces across the entire network. The experience of using blockchain systems in this fashion can be referred to as ‘Web3’. The combination of recommendation systems and blockchains will certainly have positive implications in both realms. Let’s explore how recommendation systems can improve the UX of blockchain networks, going above and beyond their impact on the Internet.
Recommendation, Onchain
Context
Recommendation models offer a transformative layer of personalization and relevance for using blockchain networks. With the onchain landscape becoming increasingly diverse, users and builders are required to engage with a variety of assets, communities, protocols and ecosystems. Sophisticated recommendation engines can dramatically reduce onchain complexities and inefficiencies, and improve UX across various levels: from wallets and protocols to DAOs and communities.
AI-powered recommendation systems are expected to assess the properties, usage patterns, and activity histories across multiple chains, generating insights for users and builders to make informed decisions.
Imagine an onchain ecosystem where every transaction, every DeFi strategy is tailored to a user’s risk appetite, every airdrop is relevant to the user’s interests, every community is aligned with the protocol’s values, every piece of content is targeted towards the most relevant consumers, and every incentive attracts the most loyal userbase. The cross-domain nature of recommendation models makes them a holistic tool for both existing and new blockchain users. It provides users a way to interact with exactly what they care about, in the most optimized way for them.
Have you ever wanted to participate in a protocol you really liked, but felt that you couldn’t get in early enough to benefit? With Web3 recommendation systems, you can be alerted to your preferred kinds of applications as soon as they launch.
Benefits
Moving beyond the simple analytics and automation tools that exist today, AI-powered recommendation models will bring the user back to the center of decision-making. Instead of supplanting human choices, recommendation models serve as a decision-optimization tool, providing suggestions and data-driven insights that complement human instincts. This approach enhances productivity and decision-making capacity while keeping the user in control.
In this regard, recommendation models embody the vision of AI and human co-evolution—AI enhances our ability to process and respond to vast amounts of information without displacing our judgment. The future of blockchain-based systems should focus on refining onchain human interactions, fostering thoughtful engagement, and personalizing experiences based on unique user preferences. Whether recommending a DeFi strategy, flagging governance opportunities, or suggesting community interactions, these models will elevate user participation and engagement to new levels.
Without human engagement, the power of Web3 ecosystems becomes limited, and focused on value-extraction. Unlocking the next phase of growth for decentralized systems can only be enabled through the symbiosis of AI’s analytical power and human intuition. Recommendation systems onchain can create environments which encourage personal interactions, human decision-making, and subjective preferences that drives innovation.
Web3 Recommendation Model Architecture
Lucidity’s approach to recommendation models is uniquely adapted to the Web3 landscape. Traditional recommendation models are built in siloed environments, focusing on a single domain like user interaction or asset management. However, in Web3, where data is decentralized and cross-domain interactions are frequent, Lucidity's recommendation models are designed to be applied across three fundamental dimensions: assets, protocols, and wallets.
-
At the protocol level, the models will assess the broader ecosystem, recommending platforms or decentralized applications (dApps) to better allocate resources and attract the right target audience.
-
At the asset level, the models will evaluate the asset history and behavior to manage the complex interplay between various assets, to eliminate inefficiencies and optimize liquidities.
-
At the wallet level, the models help provide context-specific suggestions—whether it's alerting the user to governance votes or recommending DeFi strategies based on individual risk appetite.
Lucidity’s ecosystem is envisioned to evolve into a matrix of prediction models for foundational characteristics of protocols, assets and wallets. These prediction models will be modularly fused to vectorise entities. These vector embeddings in turn will support two-way matching (or recommendations) based on demand. This multi-layered architecture ensures that recommendations are not only relevant but also dynamic to serve a variety of use cases. For more details on the architecture, check out Lucidity’s docs.
We’re at an inflection point for onchain user engagement that will be fueled by a seamless blend of AI-powered insights and personal decision-making.
Verifiable AI & Recommendation
Importance of Verifiability
In these rapidly evolving landscapes, it is important to remember the characteristics of decentralized systems which have attracted millions of users and years of R&D. Decentralization is meant to remove trust and reliance on centralized institutions to have you, the user’s, best interests at heart. Time and time again, this trust has been broken by centralized entities leading to loss of value for customers. Consequently, the importance of verifiability for AI applications onchain cannot be overstated, especially with integrations at the wallet or protocol level and closest to control of a user’s assets.
There are clear efficiency, accessibility and sophistication benefits to integrating AI into blockchain ecosystems, but to do so without verifiability risks inviting centralization back in to control and manage this global network of interactions. This would be working backwards. Without verifiability, users will be subject to value extraction by opaque AI systems that use inference manipulation to exploit their transactions and onchain decisions. This undermines the very principles of decentralization and self-custody. Verifiable AI ensures that users can trust the recommendations, knowing they align with their best interests and are free from hidden manipulation. In a decentralized system, where self-custody is paramount, the AI models integrated into wallets and protocols must be transparent, auditable, and decentralized.
Without verifiability, we risk recreating the very centralized power structures that blockchain was designed to dismantle.
Verifiability Solutions
AI can be proved verifiable by cryptographic frameworks like Zero-Knowledge ML (zkML) and Optimistic ML (opML), or by cryptoeconomic consensus systems.
ORA Protocol offers opML through its AI Oracle network and zkML through its performant library. We are excited to be working with Lucidity to bring verifiable recommendation models to users and builders across Web3 ecosystems.
About Lucidity
Lucidity is building the foundational layer for personalization in Web3. Unlike Web2, where engagement is driven through recommendation algorithms, onchain applications remain generic. Lucidity addresses the lack of personalized onchain interactions through an ensemble of prediction models fused into a general-purpose matching engine, setting the stage for the next generation of on-chain UX.
https://lucidity.finance | x.com/LucidityFinance
About ORA
ORA is chain-agnostic infrastructure that bridges the gap between AI and blockchain.
We empower developers with the tools necessary to build end-to-end trustless and decentralized applications enhanced by verifiable AI.
ORA is live today, offering the capability to verify and run inference on the largest and most sophisticated AI models without limitations.