Skip to main content

This site is for educational purposes only. Nothing here constitutes financial advice.

Topic 52 of 189

Intent-Based Architectures

Why CoW Swap, UniswapX, Across, and 1inch Fusion moved from 'sign a transaction' to 'sign an intent' — and how solver competition delivers better prices and MEV protection.

Beginner
8 min readUpdated November 2026Block Clarity Hub Editorial Team

The basics

The Old Way: You Pick the Path

Traditional DeFi swaps work like this: your wallet decides which DEX to use, which path through which pools, what the slippage tolerance should be, and submits a transaction with all of that hardcoded. If a better path emerges before your transaction lands, you don't benefit. If MEV bots see your transaction in the mempool, they can sandwich it. The user (or their wallet's router) is making every decision before submitting, then committing to it.

The New Way: You Sign What You Want

Intent-based architectures invert this. You sign a message saying 'I want at least 1.95 ETH for my 5,000 USDC, by [deadline].' The intent doesn't specify the path. Off-chain solvers compete to fulfill it — finding the best route through DEXes, bridges, AMMs, private quoter pools, even matching against opposite intents — and submit the winning execution on-chain. You get a price at least as good as your intent demanded, and often much better.

Where You Already Use It

CoW Swap has run this model in production since 2021. UniswapX (2023) brought it to Uniswap. 1inch Fusion (2022) brought it to 1inch. Across uses intent-based bridging. If you've used any of these, you've signed an intent. The user experience often looks identical to a regular swap — same wallet flow, same confirmation — but the on-chain settlement is different.

  • Old: sign a transaction with a specific route, slippage, deadline
  • New: sign an intent saying 'fulfill these conditions by this time'
  • Solvers compete off-chain to find the best execution
  • Used by: CoW Swap, UniswapX, 1inch Fusion, Across, many more

Key Takeaways

  • Intents declare what you want, not how to get it
  • Solvers compete to fulfill — competition pushes prices toward optimal
  • MEV resistance is structural: you signed a price, not a transaction in the public mempool
  • Already in production at major DEXes and bridges

Going deeper

How CoW Swap's Batch Auctions Work

CoW Swap runs a batch auction every ~12 seconds. Users sign intents during the batch period. At the end, solvers (bonded, whitelisted off-chain bidders) compute proposed settlements that fulfill all intents in the batch and submit them. Solvers can bid on individual orders or on groups of orders (a combinatorial auction); the protocol selects the combination of winning bids that maximises the surplus users receive, and the winning solvers settle on-chain. The 'CoW' (Coincidence of Wants) part: if two users want opposite trades (Alice wants ETH for USDC, Bob wants USDC for ETH), the batch can match them directly without touching a DEX at all — both get a better price than they would have on the open market, and no MEV is exposed.

UniswapX's Dutch Auction Approach

UniswapX uses a different mechanism: a request-for-quote (RFQ) step followed by an exclusive Dutch auction. The user signs an intent with auction parameters; vetted quoters bid in the RFQ, and the winner gets a short exclusivity window to fill (about 24 seconds, or 2 blocks, on Ethereum mainnet). If it does not fill, the order becomes an open Dutch auction in which the acceptable price decays and any filler can step in. The first filler to fill at a profitable price gets the trade. This produces fast settlement with solver competition, at the cost of less aggregate optimisation than batch auctions.

Solver Architectures

Solvers are sophisticated off-chain agents. They monitor: every DEX's reserves across chains, private market-maker liquidity, MEV opportunities, bridge corridors. To win consistently they need: fast access to liquidity data, capital efficiency (some solvers settle in their own inventory rather than touching DEXes), low latency to settlement venue. Solver firms (Barter, for example, has operated a CoW solver) are professional operations that capture margin from the spread between their execution price and the user's intent.

Intent Standards Are Still Forming

ERC-7683 is an emerging standard for cross-chain intents. Open Intents Framework (OIF) is another effort. The interoperability story is incomplete — UniswapX intents don't natively work with CoW Swap solvers, and vice versa. As of 15 November 2026, ERC-7683 is still listed as a Draft on eips.ethereum.org.

Key Takeaways

  • CoW Swap uses batch auctions where solvers compute aggregate settlements every ~12 seconds
  • UniswapX uses Dutch auctions where price decreases until a solver fills
  • Solver firms have become a meaningful market with capital, latency, and inventory advantages
  • Cross-protocol intent standards (ERC-7683, OIF) are still maturing

Under the hood

MEV Protection: Intents vs Encrypted Mempools

Two distinct approaches to MEV protection. Intent-based: the user never broadcasts a transaction; they sign an off-chain intent that the winning solver settles. No mempool exposure, no sandwich vulnerability. Encrypted mempool (proposed for Ethereum): users broadcast encrypted transactions; nodes commit to including them before they're decrypted. Both reduce sandwich attacks, but intents do it by removing the mempool from the flow entirely. Intent-based is shipping in production; encrypted mempools are further behind: Shutter has run one on Gnosis Chain for over two years but has announced it will decommission that deployment on 25 November 2026, and encrypted transactions for Ethereum mainnet are still at the draft-EIP stage.

Cross-Chain Intents

Across uses intents for cross-chain transfers. User signs 'I want X tokens on Chain B in exchange for Y tokens on Chain A by deadline T.' A solver pre-funds Chain B (delivering X tokens to the user), then is repaid with the user's Y tokens on Chain A once the protocol verifies the Chain B fill. The user gets cross-chain transfers in about 2 seconds, per Across's docs; the solver takes the spread for bearing the temporary capital exposure. This is operationally much faster than message-passing bridges and reduces user-visible latency from minutes to seconds.

Solver Centralisation Concerns

Several intent systems use a permissioned solver set — only approved solvers can fulfill intents (CoW Protocol solvers are bonded and whitelisted; UniswapX vets its RFQ quoters but lets anyone fill an open auction; Across runs a permissionless relayer network). Permissioning aids safety but raises decentralisation concerns. If three solvers dominate fulfillment, they could collude to extract value from users. CoW Protocol, for example, requires solvers to be bonded and can slash them, which aims to keep the competition honest. The MEV-Boost-style 'separating proposers from builders' debate is replaying in the intent space as 'separating intent authors from solvers' with similar centralisation tradeoffs.

Intent Composability

An emerging design space: composable intents. 'Buy ETH if it drops below $X' is an intent that lies dormant until conditions fire, then triggers solver fulfillment. 'Rebalance my portfolio to 50% ETH / 30% BTC / 20% USDC weekly' is an intent that recurs. This essentially recreates limit orders + structured products + DCA bots in a unified framework. CoW Swap supports limit orders this way.

Watching Solver Competition

CoW Swap and UniswapX publish settlement data showing which solver won each batch/intent. This is real-time competitive intel: which solvers consistently win, what spreads they capture, how price improvement compares across solvers. Dune dashboards by community analysts make this tractable. For users, the win rate of different intent protocols on the same trade pair is the practical question.

Key Takeaways

  • Intents bypass the public mempool — structural MEV protection without encryption
  • Across uses cross-chain intents to deliver ~2-second bridges via solver capital
  • Solver centralisation is a live concern; permissioned solver sets vs open ones are a tradeoff
  • Composable intents (limit orders, DCA, conditional triggers) are the next design wave
  • Solver competition data is public; track it to choose between intent protocols

Related Content

Hash Functions Compared

SHA-256, Keccak-256, Blake3, and Poseidon — which one each chain uses, why ZK systems needed a new family of 'arithmetic-friendly' hashes, and what tradeoffs each makes.

Stealth Addresses and Confidential Transactions

Privacy primitives that hide who's receiving what — from Monero's foundational stealth addresses to Ethereum's ERC-5564 and the legal context post-Tornado-Cash.

CDP Lifecycle

How collateralized debt positions work end-to-end — minting DAI against ETH, paying stability fees, surviving liquidations, and the structural lessons from MakerDAO's Black Thursday and Liquity's no-fee model.

Perpetual Swap Mechanics

How perp futures actually work — funding rates that keep them pegged to spot, mark vs index price, insurance funds that backstop losses, and the liquidation cascades that wreck overleveraged accounts in seconds.

MEV-Share Mechanics

How Flashbots' MEV-Share lets users share metadata about their transactions in exchange for a cut of the MEV searchers extract — a structural shift from 'searchers prey on users' to 'searchers and users share value.'

Solver Networks

How solver markets shape DEX prices, the competitive structure across CoW Swap, 1inch Fusion, Hashflow, and other venues, and why solver consolidation is a central debate in DeFi market structure.

Liquidation Cascades

The mechanical structure of cascading liquidations — Black Thursday, May 19 2021, August 2024, October 2025 — what amplifies them, who profits, and why over-leverage creates correlated risk across the entire DeFi stack.

Flash Loan Economics

Why flash loans exist, the legitimate uses (arbitrage, collateral swaps, refinancing), and how the same primitive that enables capital-efficient DeFi also enables governance attacks and oracle manipulation.

References & further reading