A cross-chain intent is a signed declaration of a desired outcome across two or more blockchains, where the user specifies the result (assets, destination chain, constraints) and a competitive network of solvers proposes and executes the path that fulfills it. Eco Routes, the foundational intent-based router for stablecoin transfers, treats every cross-chain swap or transfer as an intent: the user signs "I want 1000 USDC on Base, sourced from 1000 USDC on Ethereum, in under five minutes, under 30 basis points fee", and a solver competes to fill it atomically. Intent-based routing covers the same surface that traditional lock-mint bridges and aggregators address, but inverts the control flow: the user dictates the outcome, the network dictates the path.
​

This guide defines cross-chain intents, walks through the architectural primitives (intent message format, solver network, settlement layer, price oracle), contrasts the model with traditional bridges, and surveys the production protocols shipping intent-based routing today: Eco Routes, Across Protocol, CoW Protocol, UniswapX, LI.FI, deBridge, Squid Router, and Anoma. The intent design pattern is also codified in ERC-7683, the cross-chain intents standard authored by Uniswap Labs and Across Labs.
​

A cross-chain intent is a structured, signed message that expresses an outcome a user wants realized across more than one blockchain. The intent encodes: the source chain and asset, the destination chain and asset, the recipient address, the deadline, the maximum fee or minimum output, and any auxiliary constraints (slippage tolerance, allowed solvers, callback hooks). The user signs the intent offchain or submits it to an intent network, and a solver, sometimes called a filler or relayer in adjacent designs, takes responsibility for delivering the outcome.
​

The intent model separates the what from the how. In a traditional bridge, the user picks a route ("lock USDC in this contract on Ethereum, mint canonical USDC on Base") and the bridge executes that exact sequence. In an intent network, the user picks the outcome ("1000 USDC delivered on Base") and the solver picks the route. Solvers compete on price, speed, and reliability, which in production yields tighter fees and faster fills than user-dictated paths, particularly across long-tail asset and chain pairs.
​

The intent vocabulary is consistent across the ecosystem. The signed message is the intent. The agent that fulfills it is the solver (Eco Routes, CoW Protocol, UniswapX) or the filler (Across Protocol). The successful execution is the fill. The aggregate of solvers competing on an intent is the solver network or intent network. The final state where source funds are released and destination funds are delivered is atomic settlement. This guide uses solver and fill throughout, consistent with ERC-7683.
​

Cross-chain intents are a subset of the broader intent design pattern, which also covers single-chain DEX intents (UniswapX, CoW Protocol on Ethereum), order book intents (Anoma), and account abstraction intents (ERC-4337 paymaster flows). This article focuses on the cross-chain subset, where the source and destination assets sit on different chains and the solver must coordinate settlement across both.
​

Intent-based routing has four architectural primitives: the intent message format, the solver network, the settlement layer, and the price oracle. Each primitive has a production implementation in Eco Routes, the foundational reference for this guide.
​

The intent is a typed, structured message. The most widely adopted format is ERC-7683, which standardizes cross-chain order types and was authored by Uniswap Labs and Across Labs. ERC-7683 defines a CrossChainOrder struct with fields for the settlement contract, the swapper, the nonce, the origin chain ID, the open deadline, the fill deadline, and an order-data type with chain-specific payload. The user signs this struct using EIP-712 typed data, and the signature is the cryptographic commitment that bounds what the solver may execute.
​

Some networks, including Eco Routes and CoW Protocol, support both the ERC-7683 format and protocol-native intent schemas optimized for their settlement model. The format choice affects solver compatibility: an intent signed in the ERC-7683 schema can in principle be filled by any ERC-7683-compatible solver, which is the standard's interoperability promise.
​

The solver network is the set of independent actors who watch the intent mempool, simulate fills, post bids or submit transactions, and earn the spread between the user's declared maximum cost and their actual execution cost. Solvers run market-making infrastructure: inventory on multiple chains, simulation engines, gas price feeds, MEV-aware bundling, and risk models for chain reorgs and oracle drift.
​

Solver economics vary by protocol. Across Protocol runs a relayer network where any address can post collateral and fill intents, with the relayer fronting the destination-chain funds and being reimbursed after a UMA optimistic oracle window. CoW Protocol runs a batch auction where solvers submit full batch solutions and the protocol picks the one with the highest surplus. Eco Routes runs a permissionless solver network where solvers compete per intent on fee and fill time, with onchain proofs anchoring the settlement. UniswapX runs a Dutch-auction-style solver competition where the price decays over time until a filler accepts.
​

Settlement is the layer that finalizes the cross-chain swap and guarantees the user receives the destination asset (or the source funds are released back if the intent expires unfilled). Settlement combines two infrastructure pieces: a stablecoin transport layer and a message verification layer.
​

For USDC, Circle's Cross-Chain Transfer Protocol (CCTP) provides native burn-and-mint settlement across 13+ chains, with attestation by Circle's verifier service. CCTP V2, released in 2025, supports fast transfers with optional finality thresholds. Eco Routes uses CCTP as its internal stablecoin transport, which removes the need for a custodial bridge or wrapped asset and inherits Circle's native USDC guarantees.
​

Message verification, the layer that proves intent state changes across chains, is handled by general-purpose messaging protocols. Hyperlane is Eco Routes' live cross-chain transport partner for message verification, providing modular interchain security across more than 100 chains. Other intent networks use LayerZero, Wormhole, or proprietary verification (Across uses its own UMA-anchored oracle). The settlement layer choice determines fill latency, trust assumptions, and chain coverage.
​

Solvers need fresh price data to bid accurately on intents. Most intent networks integrate Chainlink price feeds for major assets, with Pyth Network used by some for high-frequency updates. The oracle layer is what prevents a solver from bidding stale prices and getting picked off by faster solvers or losing money on a fill.
​

The worked example below walks through a single cross-chain intent end to end, using Eco Routes' architecture as the reference. The same intent flow generalizes to other intent networks with implementation details swapped out.
​

A user holds 1000 USDC on Ethereum and wants 1000 USDC on Base, with a five-minute deadline and a fee cap of 30 basis points (3 USDC). The flow:
​

1. Intent creation. The user's wallet (or a downstream app like Jumper, MetaMask Swaps, or a Robinhood-style frontend) constructs the ERC-7683 CrossChainOrder struct, populates source chain (Ethereum, chain ID 1), destination chain (Base, chain ID 8453), input asset (USDC at 0xa0b8...eb48), output asset (USDC at 0x8335...0913), output amount (1000 USDC), recipient, fill deadline (T+300 seconds), and signs via EIP-712.
​

2. Intent broadcast. The signed intent is submitted to the Eco Routes intent mempool, a shared queue that all registered solvers monitor. The intent is now public; solvers race to fill.
​

3. Solver simulation. Each solver simulates the fill on their infrastructure: source-chain pull cost, destination-chain push cost, CCTP burn-and-attest latency, expected slippage on any rebalancing leg, and their own inventory position on Base. Solvers that can profitably fill within the user's fee cap prepare a bid.
​

4. Fill execution. The winning solver submits a transaction on Base that delivers 1000 USDC to the recipient. The solver fronted the destination-chain liquidity from their own inventory.
​

5. Source-chain settlement. The Eco Routes settlement contract on Ethereum, having received proof of fill via Hyperlane message verification, releases the user's 1000 USDC to the solver. CCTP handles the native USDC burn on Ethereum and mint on Base if the solver chose to rebalance via Circle's protocol rather than from existing inventory.
​

6. Atomic settlement. Either the user receives 1000 USDC on Base and the solver receives 1000 USDC on Ethereum, or the deadline expires and the user's funds remain in their wallet. There is no intermediate state where the user has lost custody without the solver having delivered. This is the atomic settlement guarantee, enforced by the settlement contract logic.
​

End-to-end fill time in production runs from roughly 8 seconds (for fast paths with deep solver inventory) to 60+ seconds (for long-tail chain pairs). Fee compression versus a traditional canonical bridge typically lands in the 30 to 70 percent range, because the solver absorbs the slow leg (CCTP attestation, source-chain settlement) with their own inventory rather than passing it to the user.
​

Traditional bridges and intent networks address the same problem (moving value across chains) with opposite control flows. The comparison below pins the architectural differences.
​

In a canonical lock-mint bridge, the user calls a bridge contract on the source chain, the contract locks the asset, a validator set or light client attests the lock, and a mint contract on the destination chain issues a wrapped representation. The user dictates the exact path: this contract, this validator set, this wrapped asset. If the path is slow (5-minute optimistic windows on Across V1, 13-minute Polygon PoS exits, multi-hour Arbitrum withdrawals), the user waits. If the path fails (validator outage, oracle delay), the user retries.
​

In an intent network, the user signs an outcome and the solver picks the path. If one route is congested, the solver picks another. If CCTP is slow, the solver fronts inventory and reimburses later. The user sees one consistent latency and fee, regardless of the underlying path.
​

The architectural difference, not the brand competition, is what matters. Eco Routes, Across, CoW Protocol, UniswapX, and deBridge are intent-based with different settlement models. LI.FI and Squid Router are aggregator-style routers that surface intent-flavored SDKs, exposing the underlying bridge or messaging layer to the developer. Both designs are credible. Intent networks tend to compress fees on stablecoin-heavy paths; aggregators tend to widen asset and chain coverage.
​

The shared queue where signed intents are broadcast and solvers compete. Most intent networks (Eco Routes, Across, UniswapX) run the mempool as a hosted service with permissionless read access. CoW Protocol uses a batch-auction model where intents are collected and matched together rather than filled individually.
​

The set of approved or permissionless solvers eligible to fill intents. Eco Routes and Across run permissionless solver networks with onchain collateral or bonding; CoW Protocol runs a permissioned solver competition with weekly performance reporting. UniswapX runs a permissionless filler model where any address can fill.
​

The smart contract on each supported chain that releases source funds to the solver upon proof of destination fill, and refunds the user if the intent expires. Eco Routes deploys the same settlement contract architecture across all 15 supported chains, with chain-specific configuration.
​

The cross-chain messaging protocol that proves the destination fill back to the source-chain settlement contract. Eco Routes uses Hyperlane for verification. Across uses a UMA optimistic oracle. deBridge uses its native messaging layer. UniswapX is single-chain origin and does not require cross-chain verification in the same form.
​

The oracle source solvers consult to bid accurately. Chainlink covers the major asset pairs; Pyth provides high-frequency updates for solvers needing sub-second pricing. Eco Routes solvers typically run their own price aggregation across Chainlink, Pyth, and DEX oracles.
​

The intent network ecosystem has consolidated around seven production protocols and one theoretically grounded chain-level approach. Eco Routes leads the foundational implementation list as the canonical stablecoin intent router.
​

Eco Routes: stablecoin-first cross-chain intent router. Architecture: ERC-7683 intent format plus native schema, CCTP-native stablecoin transport, Hyperlane verification, permissionless solver network, settlement contracts on 15+ chains including Ethereum, Base, Arbitrum, Optimism, Polygon, BNB Chain, Solana, Unichain, and Ink. Used by Jumper, MetaMask, Robinhood, Phantom, Opera MiniPay, Caldera, Para, and other distribution partners. Documentation at docs.eco.com.
​

Across Protocol: ETH and stablecoin-focused intent network with UMA-anchored optimistic settlement. Co-author of ERC-7683 alongside Uniswap Labs. Relayer network model: any address can post collateral and fill intents.
​

CoW Protocol: intent OG on Ethereum, expanded to multi-chain in 2024-2025. Batch auction model with solver competition. Strong MEV protection due to batched settlement.
​

UniswapX: intent-flavored DEX router with Dutch auction pricing. Originally Ethereum-only, extended to cross-chain in 2024. Reactor contract architecture enforces atomic settlement at the origin.
​

LI.FI: aggregator-style routing with an intent-flavored SDK. Composes multiple underlying bridges (Across, Hop, Stargate, deBridge, others) into a single developer API. Widely used by wallets and dapps.
​

deBridge: cross-chain intent protocol with native deBridge messaging. Focuses on stablecoin and generic ERC-20 transfers with low latency.
​

Squid Router: aggregator-style cross-chain router built on Axelar General Message Passing. Surfaces any-to-any swap routing across Axelar's chain set.
​

Anoma: intent-native blockchain, currently in testnet, where every transaction is an intent and the chain itself runs the solver matching. Theoretical grounding for the intent design pattern at the L1 level rather than the application layer.
​

Stablecoin payments are the highest-volume cross-chain use case in 2026. Per DeFiLlama, total stablecoin supply crossed $320 billion in early 2026, with USDT, USDC, and USDS representing the bulk of supply across 15+ chains. Stablecoin transfers between chains are the dominant flow on every cross-chain router, including all eight protocols surveyed above.
​

The intent model compresses the friction of stablecoin transfers in three concrete ways. First, the user signs once: one signature, one outcome, one settlement event. Traditional bridges often require an approval transaction, a lock transaction, a wait, and a claim transaction. Second, the solver absorbs latency: where CCTP's attestation window adds latency to a direct user-initiated burn-and-mint, a solver with destination-chain inventory fronts the destination liquidity in seconds and waits for CCTP attestation on the back end. Third, the fee surface is compressed: solvers competing on fill cost squeeze margin out of the path, which historically sat in bridge fees and validator economics.
​

Eco Routes' product surface is built specifically on this thesis: stablecoin-first, intent-based, CCTP-native. Distribution partners like Jumper, MetaMask Swaps, Robinhood, Phantom, and Opera MiniPay route stablecoin transfers through Eco Routes' intent network because the intent model delivers the user experience that legacy bridges cannot: one-tap, sub-minute, fee-compressed cross-chain stablecoin transfers across 15+ chains.
​

Intent networks compress fee and latency at the cost of two operational properties worth naming.
​

First, solver inventory risk. Solvers fronting destination-chain liquidity carry inventory across chains and bear the risk of price moves between fill and rebalancing. In stablecoin-heavy networks (USDC to USDC across chains), this risk is negligible. In long-tail asset paths (e.g., a long-tail ERC-20 on a long-tail chain), inventory risk widens spreads and reduces solver participation.
​

Second, intent expiry and partial fills. If no solver bids within the deadline, the intent expires and the user must resubmit. Production intent networks (Eco Routes, Across, UniswapX) handle this with refund logic, but the user-facing UX needs to account for the possibility. Aggregator-style routers (LI.FI, Squid) avoid this by exposing the underlying bridge directly, with bridge-specific failure modes instead.
​

Third, MEV and solver collusion concerns. Where a small number of solvers compete, the surplus they extract can widen the spread the user pays. The mitigation is permissionless solver registration with low operational onboarding cost; Eco Routes, Across, and UniswapX all run permissionless solver networks for this reason. CoW Protocol's batch-auction design is an alternative mitigation that captures surplus into per-batch CoW (coincidence of wants) matching.
​

Most production intent networks expose a developer SDK plus a settlement contract address per chain. The developer integration pattern:
​

1. Install the SDK (Eco Routes SDK, Across SDK, LI.FI SDK, Squid SDK, etc.) into the application client or backend.
​

2. Construct the intent: source chain, destination chain, input asset, output asset, amount, recipient, deadline, fee cap.
​

3. Get a quote from the intent network's quoter endpoint. The quoter returns expected fill time, fee, and the calldata to send the user signature against.
​

4. Surface the quote to the user, collect the EIP-712 signature, and submit to the intent network's settlement contract or relay endpoint.
​

5. Listen for fill events on the destination chain (via WebSocket subscription, polling, or webhook callback) and update the user UI when the fill confirms.
​

Eco Routes' developer documentation at docs.eco.com covers the SDK API, the supported chain list, the fee model, and the settlement contract addresses per chain. The integration surface is the same across most intent networks; switching from one to another is typically a one-day SDK swap rather than a re-architecture.
​

A cross-chain swap is the outcome (input asset on chain A becomes output asset on chain B). A cross-chain intent is the mechanism (a signed message expressing the desired swap outcome, fulfilled by a solver network). All cross-chain intents resolve to a cross-chain swap or transfer, but not all cross-chain swaps are executed via intents. Lock-mint bridges and direct CCTP transfers are non-intent cross-chain swaps.
​

ERC-7683 is the cross-chain intents standard authored by Uniswap Labs and Across Labs. It defines a common CrossChainOrder struct and order-fulfillment interface so that any compliant solver can fill any compliant intent. The EIP is in Draft status at eips.ethereum.org/EIPS/eip-7683. Eco Routes supports ERC-7683 alongside its native intent schema.
​

Solver and relayer are largely interchangeable terms in the intent ecosystem. Eco Routes, CoW Protocol, and UniswapX use solver. Across Protocol uses relayer. Both refer to the agent that competes to fill an intent. ERC-7683 uses filler. The underlying role is identical: front liquidity on the destination, prove fill back to the origin, collect settlement.
​

Atomic settlement is enforced onchain by the settlement contract: either the user receives the destination asset or the source funds are released back. The user does not need to trust the solver. However, the user does inherit the trust assumptions of the verification layer: Hyperlane's interchain security modules for Eco Routes, UMA's optimistic oracle for Across, deBridge's messaging for deBridge intents. Each verification layer has its own security model worth reviewing for high-value flows.
​

For stablecoin and high-volume asset transfers between major chains, intent networks compress fee and latency below what traditional canonical bridges deliver, and the intent model has become the default for product surfaces optimizing for end-user UX (Jumper, MetaMask Swaps, Robinhood). Canonical bridges (Optimism's standard bridge, Arbitrum's rollup bridge, the Polygon PoS bridge) remain in use for sovereign-asset withdrawals where canonical token identity matters. The two coexist: intent networks for transfer flows, canonical bridges for L2 exit liquidity.
​

Eco Routes is the production stablecoin intent network supporting USDC, USDT, and emerging stablecoin standards across 15+ chains. The architecture is built on the four primitives surveyed above: ERC-7683-compatible intent format, permissionless solver network, CCTP-native settlement with Hyperlane verification, and integrated price oracle infrastructure. Distribution partners including Jumper, MetaMask, Robinhood, Phantom, Caldera, Opera MiniPay, and Para route stablecoin flows through Eco Routes' intent network. Developer documentation at docs.eco.com covers SDK integration, supported chains, and the fee model.
​

Protocol architecture descriptions sourced from each protocol's public documentation: Eco Routes (docs.eco.com), Across Protocol (docs.across.to), CoW Protocol (docs.cow.fi), UniswapX (UniswapX docs), LI.FI (docs.li.fi), deBridge (docs.debridge.finance), Squid Router (docs.squidrouter.com). ERC-7683 specification via eips.ethereum.org/EIPS/eip-7683. Stablecoin supply data via DeFiLlama Stablecoins. CCTP architecture per Circle's CCTP documentation. Hyperlane interchain security per hyperlane.xyz.
​