ParaX Whitepaper

PDF Version

ParaX Whitepaper.pdf


The proliferation of decentralized apps (dApps) in the web3 ecosystem has been hampered by the complexity and expertise required to interact with underlying blockchain platforms. To foster more inclusive and efficient user engagement, we introduce ParaX, a cross-chain AI-powered ecosystem designed to simplify interaction with web3 applications. The ParaX ecosystem encompasses several innovative components: a Meta User Interface (MUI) that aggregates various web3 apps into an intuitive portal, account abstraction for flexible transaction formats, and a suite of mini-applications for specific task execution. Furthermore, the ecosystem employs a transaction interpreter that leverages AI to translate user intents into actionable blockchain operations, eliminating the need for users to understand the technicalities of each action. And this can only be made possible by an off-chain general-purpose provable compute layer that allows cross-domain interactions in the simplest way possible. ParaX is poised to set a new standard in web3 user experience, offering unparalleled efficiency and simplicity.

1 Introduction

In the rapidly evolving web3 ecosystem, the emergence of decentralized applications (dApps) marks an exciting leap forward. However, the prevailing design approach for most dApps involves intricate, direct interactions with the hosting blockchain. This architectural choice, while seemingly straightforward, is fraught with challenges. Given the low-level complexity of blockchain technology, users are required to possess an advanced understanding of how transactions are initiated and finalized. This technical demand has contributed to sluggish rates of adoption for web3 platforms, highlighting a pressing need for more user-friendly interfaces.
To address this challenge, we present ParaX, an innovative cross-chain ecosystem enhanced with Artificial Intelligence (AI) capabilities. ParaX aims to streamline user interaction with web3 applications by introducing an automated, smart execution layer. This system effortlessly interprets user objectives, converting them into a series of blockchain transactions that can be executed and validated with minimal user input. By significantly reducing the complexity of blockchain interactions and furthermore, reducing the cost of the transactions by utilizing provable zk-based off- chain computation layer. ParaX aspires to eliminate the barriers that have impeded the mainstream adoption of web3 applications.

The ParaX Ecosystem

ParaX offers a vertically integrated infrastructure combined with a user interface optimized for web3 applications. This system aims to cater to billions of non-cryptocurrency users, ensuring ease of use. The ParaX ecosystem comprises of:
  • Meta User Interface (MUI): A primary interface to access the underlying functions.
  • Account Abstraction: A multi-chain account system empowered by account abstraction.
  • Mini-Applications: Compact, intent-driven applications offering specific services which can be on/off chain.
  • PEPS Stack Mobile Application: Progressive Web Applications sidestepping traditional app store approvals.
  • ParaCompute: General purpose provable off-chain compute engine that hosts off-chain mini-apps.
  • AI-Intent Interpreter: An AI-driven module to translate user intents to actionable requests.
In this setup, users chiefly engage with mini-apps integrated within the MUI, rendering streamlined on-demand operations. All strategies facilitated by these mini-apps are verified via ParaCompute. Given the intricacies of cross-chain User Operations, LayerZero functions as the essential cross-chain messaging layer.

2 Meta User Interface (MUI)

In the current web3 landscape, applications tend to be highly specialized, each focusing on a single functionality. As a result, users frequently switch between multiple dApps to accomplish a range of tasks, many of which are complementary. Compounding this complexity is the fact that these dApps can reside on different blockchains or even Layer 2 solutions. This fragmented ecosystem often discourages users, pushing them towards centralized platforms that offer similar functionalities in a more streamlined and user-friendly manner.
ParaX seeks to address this disjointed user experience by establishing an intuitive, omni-chain portal that aggregates a myriad of web3 applications. Our solution effectively creates a unified interface, serving as a single point of entry for various blockchain-based services and utilities.
To realize this vision, ParaX introduces a permission-less Meta User Interface (MUI) empowered by a Software Development Kit (SDK). This SDK allows other dApp developers to seamlessly integrate their applications into the ParaX ecosystem, promoting a cohesive user experience across all partnered platforms. Additionally, our MUI fosters interoperable communications between apps, setting the stage for action-oriented mini-apps, a topic we will delve into in subsequent sections of this whitepaper.
Figure 1: Unified interface for all user interactions
To kickstart ParaX MUI, ParaX team will be developing and integrating a list of the most used apps in the ecosystem. This will help users onboard to the ecosystem to achieve their most fundamental needs.

2.1 Progress Web Apps integrations for mobile

With the combination of MUI and Progressive web applications concepts, The apps integrated on ParaX can be mobile compatible. This can remove the overhead of submitting an app for approval on traditional Apple and Android Stores without compromising the performance of the apps and achieving the goals of MUIs on mobile apps.
Figure 2: MUI - All user needs in one place

3 Account Abstraction

In recent years, the blockchain space has witnessed a plethora of innovations aimed at improving both scalability and user experience. Among these developments, one stands out due to its potential for wide-ranging impacts on how users interact with blockchain systems: account abstraction.
At its core, account abstraction is an advanced mechanism that uncouples the logic of user account operations from the rigid constraints and predefined rules of the blockchain platform. In simpler terms, it offers a layer of abstraction that allows for flexible transaction formats and broadens the decision criteria for transaction validity.

3.0.1 The Importance of Account Abstraction

Account abstraction addresses multiple issues prevalent in traditional blockchain frameworks. Firstly, it paves the way for innovative features, allowing developers to introduce advanced functionalities without being tethered to the fixed rules of the underlying blockchain. Secondly, by decoupling user operations from specific blockchain rules, it ensures broader compatibility and adaptability, making the technology future-proof. We will be discussing some of these benefits in the following sections
1. Automation of User Actions
One of the paramount advantages of account abstraction is the automation of user-centric actions. By utilizing abstracted frameworks integrated with smart contracts, actions such as recurring payments, stake validations, and automated trades can be pre-configured. This not only reduces the manual intervention required by users but also minimizes human-induced errors, thereby fostering a smoother and more efficient transactional experience on the blockchain.
2. Sponsoring User Transactions
Account abstraction has the potential to redefine the economic model of transactions. Traditionally, users have to pay for transaction fees, known as ’gas.’ However, with the abstraction framework in place, third-party entities or sponsors can be programmed to subsidize these costs. Such a model can catalyze increased user participation by reducing transactional expenses and also provides avenues for businesses to incentivize user interactions or engagements.
3. Facilitating Cross-chain Actions
Perhaps one of the most compelling potentials of account abstraction lies in its ability to bridge actions across different blockchain networks. By abstracting account operations from the inherent intricacies of individual chains, it becomes feasible to execute transactions or transfer data across multiple chains. This interoperability is critical for a decentralized future, where multiple blockchains coexist. Through account abstraction, seamless cross-chain actions can be realized, reducing the friction and barriers that currently exist between different blockchain ecosystems.

3.0.2 Account Abstraction within ParaX

Account Abstraction serves as a foundational pillar in ParaX’s ecosystem, significantly mitigating the limitations inherent in Externally Owned Accounts (EOA) that typically require user involvement at each interaction point. This feature simplifies the multiple steps traditionally needed to achieve a singular objective, thus reducing the overhead and complexity associated with blockchain roles.

3.0.3 Case Example: NFT Liquidation and Borrowing Procedures

To elucidate the utility of Account Abstraction, let’s consider a hypothetical user aiming to liquidate an NFT asset to initiate a borrowing action on a Layer 2 solution like Arbitrum. In the conventional model, this user would navigate through the following steps:
  1. 1.
    Approve NFT Collection
  2. 2.
    Execute NFT Sale
  3. 3.
    Withdraw ETH from Blur
  4. 4.
    Bridge ETH to Arbitrum via Stargate
  5. 5.
    Authorize ETH on Arbitrum’s AAVE
  6. 6.
    Supply ETH
  7. 7.
    Borrow ARB Token
Under optimal conditions, executing these steps can take a minimum of 90 seconds, or even stretch up to 10 minutes, primarily depending on the time required for blockchain bridging.
Efficiency Gains with ParaX: In contrast, utilizing ParaX’s Account Abstraction, steps 1-3 and 5-7 can be aggregated into individual transactions. This optimized approach dramatically cuts down the transaction time to approximately 30 seconds for non-bridging operations while also achieving an estimated 20% reduction in gas costs.

3.0.4 Leveraging ERC-4337 for Advanced Account Abstraction

The ERC-4337 [2] standard has revolutionized the concept of Account Abstraction (AA), offering a framework for efficient and modular implementations. ParaX capitalizes on this standard to streamline AA, enabling a highly user- friendly experience. With this approach, users simply need to indicate their intentions by signing a message that outlines their desired actions.
Upon receiving this signed message, the Transaction Interpreter deciphers the user’s intent and routes it to one of the available bundlers for execution. Should the transaction require sponsorship for its cost, the PayMaster component seamlessly handles this aspect.
As a result, users engage with a singular entity, sparing them the need to understand the intricacies and rules associated with blockchain operations. This simplifies the user experience substantially, making it more accessible for individuals regardless of their familiarity with blockchain technology.
Figure 3: Account Abstraction Flow

4 Intent-based Transaction Interpreter

This element is integral to the ParaX ecosystem, responsible for analyzing and directing transactions based on discerned user intentions.
What is intent? In conventional transaction procedures, a validator, upon receiving a transaction signature, is obligated to adhere to a predetermined computational route based on a specific state. A fee also acts as a motivator for the validator to act accordingly. In contrast, intents operate differently. Rather than dictating a set computational course, an intent permits any route that satisfies certain criteria. When a user endorses and disseminates an intent, they empower the recipients to select an appropriate computational route for them. This unique feature allows for a more exact characterization of intents as authenticated messages that enable a sequence of state changes, starting from an initial point. [3]
Figure 4: Intent Interpreter
Using State Space: A state space approach is ideal for formalizing intents in blockchain networks. State space provides a mathematical framework that allows us to represent all possible states of a system and the transitions between them. In the context of blockchain networks like Ethereum, each state can represent a specific configuration of a user’s assets, smart contracts, and other blockchain-related elements. The transitions between states can represent transactions, contract executions, or any other actions that change the state.
As you can see, the state diagram is actually a graph, where nodes are states, and edges, in the literature called transitions, are inputs that we feed to the automaton (in our case in the form of calldata) to get to another state.
Lagrangian Mechanics: The concept of Lagrangianscan be a way to evaluate the "best" path to achieve an intent. In this framework, each path has an associated "action," and the path that minimizes this action is considered the best. This is analogous to how particles in physics take the path that minimizes the action, known as the principle of least action. [1]
Solver’s Perspective: The solver is the entity responsible for finding the best path to fulfill an intent. The solver uses the Lagrangian to evaluate different paths and chooses the one that minimizes the action. This can be the job of the intent interpreter.

Types of Lagrangians for Intent Solvers

Free Solver Lagrangian (Lsf): The Free Solver Lagrangian is designed to model a solver that is solely focused on minimizing the cost of transitions, often represented by gas fees in the Ethereum network. The mathematical representation is as follows:

Lsf(q,q)=Us(q)Lsf (q, q‘) = Us(q‘)

Here, Us(q‘) denotes the utility, usually negative, associated with the transition q‘ .
Greedy Solver Lagrangian (Lsg): The Greedy Solver Lagrangian aims to model a solver that is focused on maximizing its utility by considering the utility of the states it passes through. The formal equation is:

Lsg(q,q)=Us(q)Us(q)Lsg(q, q‘) = Us(q) − Us(q′)

In this equation, q′ is the state reached after applying the transition q‘ to the state q.
Weighted Greedy Lagrangian (Lsw): The Weighted Greedy Lagrangian is a hybrid model that takes into account both the cost of transitions and the utility of states. It is mathematically represented as:

Lsw(q,q)=Us(q)+(Us(q)Us(q))Lsw(q, q‘) = Us(q‘) + (Us(q) − Us(q′))

Or equivalently,

Lsw=Lsf+LsgLsw = Lsf + Lsg

By utilizing these three types of Lagrangians, we have a formalized framework for modeling and solving the problem of intent fulfillment in blockchain networks. This approach allows for a more nuanced understanding of user objectives and offers a pathway for optimizing transactions in a cost-effective and utility-maximizing manner.

5 ParaCompute: ZK-Enhanced Off-Chain Computation Framework

In the evolving blockchain ecosystem, users frequently engage in intricate transactions spanning multiple chains. While off-chain executors can theoretically handle these transactions, the absence of execution guarantees poses challenges. Addressing this, ParaCompute introduces Zero Knowledge (ZK) proofs, enabling executors to furnish verifiable evidence of accurate off-chain computations. A salient application is the universal liquidity provisioning in DeFi protocols, where liquidity, originating from a source blockchain, disseminates across multiple chains, driven by factors like targeted APY or maximizing overall yield.

Core Features of ParaCompute:

Universal ZK Integration: ParaCompute, a versatile zero-knowledge proof network, facilitates seamless ZK proof integration across Ethereum, L1 blockchains, Cosmos app chains, L2 rollups, and dApps, expediting integration with minimal developmental overhead.
General Purpose through RISC-0/V: Rooted in the open RISC-0/V instruction set, our general purpose VM allows extensive language compatibility, supporting Rust, C++, Solidity, Go, among others. Leveraging recursive proofs, a custom circuit compiler, and state continuations, it empowers users to craft efficient zero-knowledge proofs for diverse applications.

Triad of Innovations:

  • A versatile VM, executing any virtual machine within a ZK-verifiable context.
  • An integrated proving system, compatible with any smart contract or blockchain.
  • A universal rollup, broadcasting computations verified on ParaCompute across all chains.
Web3 Ecosystem Integration: Acting as an autonomous ZK proving entity, ParaCompute enables dApps to generate ZK proofs. Once crafted, these proofs are anchored onto blockchains like Ethereum, ZK L2s, app chains, and L1s. This framework permits off-chain computations, with ParaCompute’s ZK proofs validating state changes and smart contract executions on-chain, resulting in leaner, more efficient smart contracts.
Interoperability & Universal Rollup: ParaCompute’s architecture fosters interoperability among diverse smart contracts, irrespective of their native chain. Its universal rollup capability amalgamates ZK proofs from chains like Ethereum into a consolidated proof, broadcasted across any blockchain, positioning ParaCompute as a universally accessible execution layer.
Illustrative Use Case: Consider a scenario where a user aspires to provision liquidity on both ETH and SOL, targeting the apex lending/supplying APY. The process unfolds as:
  1. 1.
    User dispatches their provisioning directive to the ParaCompute RPC node.
  2. 2.
    The relay channels this request to an executor.
  3. 3.
    The executor activates a mini-app, determining liquidity distribution across chains.
  4. 4.
    Utilizing LayerZero for cross-chain dialogue, the executor processes the request.
  5. 5.
    Execution proof is crafted.
  6. 6.
    This proof is subsequently rolled up to a designated chain via the prover relay and contract.

5.1 ParaCompute Architecture

Figure 5: ParaCompute Architecture
Proof Engine: At its core, ParaCompute harnesses the RISC Zero ZKP to validate mini-apps programs. The zkVM offers recursive proofs, a tailored circuit compiler, state continuations, and iterative algorithmic enhancements.
Proof Relay: Bridging off-chain proofs with on-chain components, the Proof Relay links dApps to ParaCompute’s off-chain modules. Smart contracts can liaise with the proof contract, soliciting off-chain executions or proofs.
RPC Node: Users retain the option to directly transmit execution requests via the RPC node, which then steers these directives to off-chain executors for processing and proof generation.

6 The Synergy of Components - Mini Apps within ParaX

By fusing the principles of Account Abstraction, Meta User Interfaces, and Automated Intent-Based Transaction Execution, and off-chain zk-based general computation setup. We unlock a robust infrastructure for mini apps. This enables developers to seamlessly integrate features from various web3 applications without reinventing the wheel for each functionality. The end result is a unified, intuitive user experience that minimizes the complexities often associated with navigating multiple apps to achieve a singular objective. This not only enhances user satisfaction but also contributes to a safer, more efficient interaction within the web3 ecosystem.
In the ParaX ecosystem, the user’s intent serves as the starting point for a sequence of actions aimed to fulfill that intent. Each action in this sequence is structured with a primary plan (Plan A), an alternative plan (Plan B), and a revert procedure for undoing the action if needed. Here is how the flow works:
Figure 6: Mini-Apps Simple Flow

6.0.1 Mini-App Action Flow

Figure 7: Mini-Apps Detailed Flow
  1. 1.
    Intent: User’s high-level goal or task.
  2. 2.
    Intent Interpreter: Interprets the intent and routes it to the appropriate registered app.
  3. 3.
    Registered App: The app responsible for the specific functionality generates the execution path.
  4. 4.
    Path Verification: zkVM (Zero-Knowledge Virtual Machine) verifies the generated path for correctness and integrity.
  5. 5.
    Execution Engine: Executes the verified path.
  6. 6.
    Proof of Execution: A cryptographic proof confirming the successful execution is sent.
Example: Suppose a user wants to withdraw ETH from the mainnet and supply it on Arbitrum. The actions could be designed as follows:
Action 1 - Withdraw.
Plan A: Withdraw ETH from the mainnet.
Plan B: N/A.
Revert Procedure: Abort the transaction.
Action 2 - Supply.
Plan A: Supply the withdrawn ETH on Arbitrum.
Plan B: Supply the withdrawn ETH on Polygon.
Revert Procedure: Resupply the ETH back to the mainnet.
In this example, the intent ("withdraw on ETH from mainnet → supply on Arbitrum") goes through the aforementioned flow. The Transaction Interpreter identifies the registered app capable of handling ETH withdrawal and supply functionalities. The app then crafts an execution path, which is verified by Risc-0 ZK Engine. Once verified, the execution engine proceeds to perform the actions, first trying Plan A and resorting to Plan B only if necessary. Finally, a proof of successful execution is generated and can be rolled up to the main chain.
This structure not only automates complex multi-step procedures but also provides a fallback and revert mechanism, enhancing reliability and user trust in the ParaX ecosystem.

7 Enhancing Transaction Interpreter with Large Language Model AI

Building upon the foundational components of the ParaX ecosystem, we introduce a groundbreaking innovation: the integration of a large language model AI into the intent interpretation process. This AI-driven approach elevates the Intent-based Transaction Interpreter, enabling it to not only decipher user intents with greater precision but also to proactively assist users in formulating the correct objects and constraints for their desired outcomes.

7.1 Key Features

Deep Intent Understanding: The AI model delves deep into the nuances of user intents, extracting subtle details that might be overlooked by traditional interpreters. This ensures that the user’s true objective is captured and executed upon.
Cross-Chain Intent Resolution: The AI is trained on a vast dataset spanning multiple blockchains, enabling it to navigate the intricacies of cross-chain operations with ease. Whether a user’s intent involves Ethereum, Binance Smart Chain, or any other blockchain, the AI can seamlessly coordinate actions across these diverse platforms.
Higher Preference Space Exploration: By leveraging AI, the interpreter can explore a broader preference space, identifying optimal paths for intent execution that might not be immediately apparent. This ensures that users get the best possible outcomes for their intents.
Proactive User Assistance: Beyond mere interpretation, the AI can actively assist users in formulating their intents. By suggesting potential objects, constraints, and even optimal pathways, the AI ensures that users can articulate their objectives with clarity and precision.
Integration with Existing Components: This AI-enhanced interpreter seamlessly integrates with existing ParaX components, such as the Meta User Interface (MUI) and Account Abstraction, ensuring a cohesive user experience.

7.2 Benefits

Enhanced User Experience: Users no longer need to grapple with the complexities of formulating cross-chain intents. The AI provides guidance every step of the way, ensuring a smooth and intuitive experience.
Efficiency: By identifying optimal pathways for intent execution, the AI ensures that actions are carried out in the most cost-effective and timely manner.
Reliability: With the AI’s deep understanding of cross-chain operations, the chances of errors or misinterpretations are significantly reduced.
Future-Proofing: As the blockchain landscape continues to evolve, the AI can be continually trained on new data, ensuring that it remains at the cutting edge of intent interpretation.

7.2.1 Fine Tuning the LLM Model

Figure 8: LLM Fine Tuning
Fine-tuning the Intent Interpreter involves a meticulous process to enhance the pre-trained Large Language Model (LLM). At the heart of this refinement is the integration of a reward model, which provides invaluable feedback, ensuring the LLM’s responses align closely with user expectations in the blockchain and web3 domain. Once the LLM is fine-tuned, it adeptly deciphers user inputs, encoding them into precise intents. This refined understanding then facilitates the generation of an optimal execution path, streamlining the user’s cross-chain operations. Key components of this process include:
  • Pre-trained LLM: The foundation upon which domain-specific knowledge is layered. - Domain-Specific Dataset: Curated data that encompasses a wide spectrum of blockchain-related intents, ensuring the LLM’s expertise in the domain.
  • Reward Model: A feedback mechanism that guides the LLM’s fine-tuning, reinforcing accurate interpretations and rectifying misalignments.
  • Evaluation Metrics: Tools like accuracy and precision that gauge the model’s performance, ensuring its reliability.
  • Encoded Intents: The LLM’s output post-fine-tuning, which translates user inputs into actionable intents.
  • Optimal Execution Path Generation: The culmination of the process, where the refined LLM determines the most efficient route to realize a user’s intent within the ParaX ecosystem.
Through this structured approach, the Intent Interpreter becomes a powerful tool, adeptly navigating the complexities of the web3 landscape to deliver seamless user experiences.

8 Conclusion

Throughout this whitepaper, we have delved into ParaX’s innovative approach to enhancing the web3 user experience. Central to our strategy is the AI-enhanced intent-based user experience, which intuitively deciphers and addresses user needs, eliminating the complexities traditionally associated with web3 interactions. By integrating this with the pioneering concepts of meta interfaces and account abstraction, we offer users a streamlined and intuitive interface. Fur- thermore, our off-chain general-purpose provable compute layer ensures efficiency, reducing the overheads commonly associated with blockchain operations. In essence, ParaX’s holistic approach is set to redefine the web3 landscape, ensuring users can navigate the ecosystem with ease, efficiency, and confidence.


  1. 1.
    Fabrizio Genovese. Lagrangian mechanics of intent solving i, 2023.
  2. 2.
    Dror Tirosh (@drortirosh) Shahaf Nacson (@shahafn) Alex Forshtat (@forshtat) Kristof Gazso (@kristofgazso) Tjaden Hess (@tjade273) Vitalik Buterin (@vbuterin), Yoav Weiss (@yoavw). Erc-4337: Account abstraction using alt mempool, 2021.
  3. 3.
    Bastian Wetzel. Intent-based architectures and projects experimenting with them, 2023.