Intent Architecture: Potential Solutions to the Complexity Issues of DeFi

Before the collapse of Luna, I ran a stablecoin yield strategy for a friend to take advantage of the exceptionally high interest rates at that time. My friend was not well-versed in cryptocurrency and had never interacted on-chain before. Our collaboration model was simple: he would deposit funds into a hardware wallet, and then we would meet once or twice a week via Zoom, where I would gradually guide him through the process.

We diversify our funds across almost all available DeFi protocols on various chains. During a 2-4 hour session, we conduct dozens of transactions, including approvals, transfers, swaps, deposits, claims, and withdrawals. Funds are allocated to various projects, such as customized Uniswap LP trading pairs and Curve vote locking, to maximize returns. We utilize almost all mainstream cross-chain bridges, DEXs, and yield aggregators to manage our stablecoin portfolio.

The most challenging part of this process is explaining all the necessary steps to my friends. I issue instructions, and they need to execute and understand the interfaces of various complex Decentralized Finance tools. Our meetings are filled with directives like "click here," "go there," and "swap this." For example, the process of exchanging USDC for FRAX/DAI LP on Polygon requires 12 transactions! This process is not only cumbersome but also time-consuming and arduous, especially when managing a large-scale portfolio.

From a macro perspective, all the processes we execute have clear expected outcomes. We have assets and wish to use them to accomplish specific tasks. For example, "We have USDC (on Ethereum), and we want to provide liquidity in the form of FRAX/DAI (on Polygon), and then deposit it into the staking vault." This is the "content" of our operation, while the 12 transactions we must execute represent the specific "how" of the operation. A series of clear and logical steps are needed from start to finish, and these steps are all quantifiable.

If there are powerful algorithms to handle transaction routing, the entire process could be completed in just 1-2 steps. Users only need to express the results they want, and the algorithm can return the best path, potentially even executing the transaction directly. This path mapping structure is referred to as "intention," and it is part of the future of middleware in the rapidly evolving Ethereum ecosystem.

Currently, there is no consensus in the industry on the definition of "intention", but there are some common viewpoints. Paradigm defines it as: "Intention is a set of declarative constraints signed, allowing users to outsource transaction creation to third parties while maintaining full control over the transaction party." David Ma from Near defines it as: "Transactions are imperative, while intentions are declarative. In other words, a transaction is a clearly defined message that specifies how to run the EVM to produce state changes, while an intention specifies the desired state changes without focusing on the specific implementation process."

Both definitions emphasize the "declarative" nature of intent, which seeks external assistance through data sharing between the user and the "solver". Users declare the outcomes they desire, and the solver provides the means to achieve them. Unlike transactions with specific parameters, intent needs to be mapped by a third party. Additionally, there are constraints that limit the possible set of paths, helping to narrow the total possibilities down to a more manageable set.

The basic architecture based on "intention" already exists in the EVM. For example, when you use a DEX, it automatically finds the best route for executing the trade. In the Curve interface, after selecting the assets to buy and sell, the UI automatically finds the best LP for routing. If there is no direct trading pair, the order will go through multiple LPs to achieve the best execution path, all completed in one transaction.

In addition to trading aggregators, there are several other types of "intents" on Ethereum, including limit orders, CowSwap-style auctions, gas sponsorship, delegation, transaction batching, and cross-chain swaps.

However, intent-based architecture also faces some challenges. The first is the issue of MEV (Maximum Extractable Value). Solvers may exploit user intents for profit, similar to payment for order flow (PFOF) in traditional finance. Secondly, the risks of middleware cannot be ignored. If a centralized Interpool absorbs most intent-based transactions, it may begin to introduce additional fees or other rent-seeking behaviors.

Nevertheless, the intent architecture still represents an important development direction for the future of Decentralized Finance. It has the potential to greatly simplify user operations, improve efficiency, and pave the way for broader adoption. Currently, some protocols are developing intent-based infrastructures, such as Flashbot's SUAVE and Anom. These developments aim to make cryptocurrency interactions more user-friendly and may open new possibilities for future blockchain applications.