Intent-Centric Architecture

Applications that abstract complex on-chain actions into simple user declarations.

• Wallet Transaction Bundling: A user signs a single intent to "swap ETH for USDC and deposit it into a lending pool," which the wallet's solver network decomposes and executes atomically.
• Gas Sponsorship: Users approve transactions without holding the native token, declaring an intent to "pay fees in USDC." A relayer fulfills this by paying gas and deducting the equivalent USDC.
• Limit Orders: Submitting an intent to "buy 1 ETH if its price falls below $3,000" is a declarative order that off-chain solvers watch and fulfill when conditions are met.

Competitive off-chain networks that specialize in fulfilling user intents efficiently and profitably.

• On-Chain Auctions: Solvers (like professional market makers or MEV searchers) compete in a Dutch auction or sealed-bid auction to fulfill a user's swap intent, providing the best exchange rate.
• Cross-Domain Intents: Solvers find optimal routing across multiple blockchains (e.g., Ethereum, Arbitrum, Polygon) to fulfill a user's intent to "bridge and swap assets with minimal cost."
• Composability: A solver might bundle hundreds of similar user intents (e.g., "sell ETH for USDC") into a single, more efficient CoW Swap batch auction to improve pricing for all.

A foundational infrastructure that enables intent expression by separating transaction validation from execution.

• UserOperations: Instead of a standard transaction, users submit a UserOperation object to a mempool, declaring their desired outcome (the intent).
• Bundlers: Specialized nodes act as solvers, collecting UserOperations, estimating gas, and bundling them into a single on-chain transaction for execution.
• Paymasters: External contracts that can sponsor transaction fees, allowing users to fulfill intents like "pay for this transaction in ERC-20 tokens" instead of ETH.

Formal languages designed to express user intents in a structured, machine-readable format for secure interpretation by solvers.

• Anoma's Taiga: A zk-circuit DSL where intents are expressed as predicates over states, enabling private execution. A user can express an intent to "trade asset A for asset B if the price is within range X," with the proof of valid fulfillment settled on-chain.
• Essential's @eth-optimism/intents SDK: Provides developers with a JavaScript library to construct and submit standardized intent objects (e.g., CrossChainSwapIntent) to solver networks.
• Cairo/Starknet: Smart contracts written in Cairo can be seen as intent fulfillment contracts, where the logic encodes the conditions for a valid state transition.

Intents that specify outcomes involving multiple blockchain networks, abstracting away the complexity of bridging and interoperability.

• Unified Liquidity Access: A user submits an intent to "borrow USDC at the lowest available rate," and a solver sources the loan from the optimal lending market across Ethereum, Avalanche, and Base, handling all cross-chain messaging.
• Intent-Based Bridges: Instead of a user manually locking tokens on one chain and minting on another, they express an intent to "move 100 USDC to Arbitrum." A solver's liquidity network fulfills this by providing the tokens on the destination chain immediately (atomic swap) and later reconciling the ledger.
• Chainlink CCIP & Axelar: These cross-chain messaging protocols can act as verifiable fulfillment layers for intents, providing the proof that conditions on another chain were met.

Complex, multi-step financial strategies expressed as a single, atomic intent, eliminating sandwich attacks and reducing slippage.

• CoW Swap & Batch Auctions: Users submit intents to trade tokens. The protocol aggregates these intents into a batch and runs a periodic auction where solvers compute the clearing price that maximizes surplus for all participants.
• Flash Loan Integration: A sophisticated user (or a solver acting on their behalf) can express an intent that logically requires a flash loan, such as "arbitrage between DEX A and DEX B." The fulfillment logic automatically takes the flash loan, executes the trades, repays the loan, and delivers the profit—all within a single atomic settlement.
• Limit Order DEXs (e.g., UniswapX): Users submit signed, off-chain orders (intents) to sell token X for token Y at a specific price. Professional market makers (solvers) fulfill these orders when the market price crosses the threshold, often providing better pricing than the AMM pool.

Users delegate execution to solvers (or fillers), creating a principal-agent problem. Security depends on the solver's honesty and competence. Risks include:

• Malicious solvers front-running, sandwiching, or stealing funds.
• Solver cartels forming to extract maximal value, reducing competition.
• Centralization pressure where a few dominant solvers control most flow, creating systemic risk.

The security of the outcome hinges on the precision of the intent expression. Vague or poorly defined intents can be exploited:

• Interpretation attacks: A solver fulfills the literal, minimal technical requirement of an intent but not the user's actual economic goal.
• Oracle dependency: Intents relying on external data (e.g., 'swap when price is X') introduce oracle manipulation risks.
• Expressiveness vs. Verifiability: More complex intents are harder for users to formally verify before signing.

Post-execution verification is critical for trust minimization. Key mechanisms include:

• State diffs: Solvers may submit proofs showing the blockchain state changed as specified.
• ZK-proofs of fulfillment: Using zero-knowledge proofs to cryptographically verify intent conditions were met without revealing solver strategy.
• Dispute resolution: Optimistic systems allow challenges within a time window, requiring bonded stakes from solvers.

Broadcasting an intent can leak sensitive information to the network, creating security risks:

• Front-running: Solvers or MEV bots can see the intent and extract value by trading ahead of it.
• Strategy exposure: Revealing a large, complex trading strategy allows competitors to counteract it.
• Privacy-preserving intents using commit-reveal schemes or secure enclaves are an active area of research to mitigate this.

The user's wallet becomes a critical trust anchor, as it signs the intent declaration. New risks emerge:

• Intent phishing: Malicious dApps trick users into signing broad, harmful intents (e.g., unlimited token approvals).
• Signer centralization: Reliance on a few intent standard implementations creates single points of failure.
• Session keys: Systems using session keys for automated intent fulfillment must carefully manage key scope and expiration.

As intent-based systems become infrastructure, they introduce new systemic risks:

• Solver failure: A major solver going offline or being compromised could disrupt entire application layers.
• Intent dependency: Smart contracts or other intents that depend on the outcome of a prior intent create complex, fragile dependency chains.
• MEV transformation: Intent architectures don't eliminate Maximal Extractable Value (MEV); they often shift it from searchers to solvers, potentially concentrating it.

A declarative transaction is a user-signed message that specifies a desired outcome (e.g., "swap X for Y at the best rate") without defining the exact execution path. This is the core data structure of intent-centric systems, shifting the burden of pathfinding and execution complexity from the user to specialized solvers or fillers.

Solvers (or fillers) are specialized, often competitive, actors who fulfill user intents. They:

• Compete to find the optimal execution path.
• Bundle transactions to capture Maximal Extractable Value (MEV).
• Are incentivized by earning fees or capturing arbitrage. This creates a marketplace where execution quality is optimized, but also introduces design challenges around solver decentralization and censorship resistance.

Account Abstraction separates a user's verification logic from their transaction execution, enabling smart contract wallets. This is a critical enabler for intent-centric UX, allowing for:

• Sponsored transactions (gasless onboarding).
• Session keys for batch approvals.
• Custom security logic (social recovery, multi-sig). It allows intents to be expressed and managed by flexible smart accounts rather than rigid Externally Owned Accounts (EOAs).

User intents often involve assets or actions across multiple blockchains. Cross-chain bridges and interoperability protocols are essential infrastructure for fulfilling these intents. Key mechanisms include:

• Liquidity bridges for asset transfers.
• General message passing (e.g., CCIP, LayerZero) for arbitrary data and calls.
• Atomic swap protocols that enable cross-chain intents to be settled trustlessly.

Traditional Automated Market Makers (AMMs) require users to specify exact parameters (pool, slippage). Intent-centric systems can abstract this by querying multiple liquidity sources. Decentralized Order Books (e.g., limit orders) are a primitive form of intent, where a user declares a price target. Modern intent architectures aggregate across both AMM pools and order books to find the best execution.

After a solver proposes a solution, the system must verify the execution correctly fulfills the intent and then settle it on-chain. This involves:

• Intent standard schemas for machine-readable constraints.
• Cryptographic attestations or zero-knowledge proofs of correct fulfillment.
• A secure settlement layer (often an L1 or L2) that acts as the final arbiter and escrow agent for involved assets.

Intent-centric systems are built on several key layers:

• User Abstraction: Wallets and interfaces that allow users to express desired outcomes in natural language or simple parameters.
• Solver Networks: A competitive network of off-chain actors (solvers) that compute and propose optimal transaction paths to fulfill the intent.
• Execution Layer: The on-chain settlement system that finalizes the solver's proposed transaction bundle, often using mechanisms like MEV auctions or batch auctions to ensure fairness and efficiency.

Leading projects pioneering intent-based design:

• UniswapX: A permissionless, auction-based protocol for trading across AMMs and private liquidity, where users submit intents filled by competing fillers.
• CowSwap & CoW Protocol: Uses batch auctions and Coincidence of Wants (CoW) to settle intents peer-to-peer or via solvers, minimizing MEV extraction.
• Anoma & SUAVE: Research-focused projects building generalized intent-centric architectures and shared sequencers for cross-domain intent settlement.

Intent-centric architecture is deeply intertwined with Maximal Extractable Value (MEV) and novel auction mechanisms. It aims to transform MEV from a negative externality into a quantifiable, contestable resource. Key mechanisms include:

• Batch Auctions: Multiple intents are settled simultaneously at a uniform clearing price, reducing front-running.
• Solver Competition: Solvers bid for the right to fulfill an intent, with the most efficient solution (often sharing surplus with the user) winning the auction.

The end-goal of intent-centric design is radical simplification for the end-user. This involves:

• Declarative Transactions: Users specify what they want (e.g., "best price for 1 ETH in USDC"), not how to achieve it.
• Account Abstraction (ERC-4337): Enables sponsored transactions, session keys, and complex logic, serving as a key enabling layer for intent expression and fulfillment.
• Cross-Chain Intents: Systems that allow a single intent to trigger actions across multiple blockchains seamlessly.

The security and liveness of an intent system depend on its solver network. Critical considerations include:

• Bonding & Slashing: Solvers often post bonds that can be slashed for malicious behavior (e.g., not fulfilling winning bids).
• Decentralization: Preventing solver cartels is essential for maintaining competitive pricing and censorship resistance.
• Verifiability: Ensuring the on-chain execution matches the user's signed intent, typically enforced by a verification contract.