As the decentralized finance ecosystem continues to expand, users are facing an increasingly complex on-chain environment. Even a simple asset swap may involve multiple liquidity pools, several protocols, and interactions across different blockchains. For everyday users, understanding these underlying mechanics and finding the best execution path is far from easy.
As an important part of DeFAI infrastructure, Velvet uses Intent-Based Trading as a core product architecture. This mechanism changes the traditional model of on-chain trading, allowing users to manage assets and execute trades without needing to understand every detail of the execution process.
Intent-Based Trading is a trading model built around “intent” rather than “instructions.” In traditional DeFi trading, users are usually required to specify the execution steps themselves, such as choosing a trading platform, setting slippage parameters, determining the trading route, and confirming gas costs. Intent-Based Trading, by contrast, focuses on the final result the user wants to achieve.
For example, a user may want to convert ETH into stablecoins, rebalance a portfolio, gain exposure to a certain asset, or execute a specific yield strategy. Under this model, users do not need to create a detailed execution plan. They only need to state their goal, and the system automatically finds the best way to achieve it.

Velvet’s Intent-Based Trading architecture is mainly made up of the user layer, the AI Agent layer, the Solver network, and the execution layer. Each module has a distinct role, and together they complete the full process from user request to on-chain execution.
The user layer receives the user’s goal rather than a specific trading instruction. For instance, a user can directly express the desire to build a certain asset portfolio or gain specific market exposure. The system then converts these requests into standardized trading intents.
The AI Agent layer interprets the user’s request and turns natural language or simple operation requests into executable on-chain tasks. At the same time, the AI Agent analyzes potential execution plans by taking into account market data, asset status, and liquidity conditions.
The Solver network is responsible for finding the best execution path. Multiple Solvers evaluate different protocols and liquidity sources at the same time, comparing prices, transaction costs, and execution efficiency before generating competing solutions.
The execution layer submits the final solution to the blockchain to complete the trade, then returns the result to the user. The entire process happens automatically in the background, so users do not need to deal with complex technical details.

After a user submits a trading goal in Velvet, the system first generates a standardized trading intent. For example, if a user wants to convert part of their stablecoin holdings into an AI-themed asset portfolio, the system identifies that goal without requiring the user to specify exactly which assets to buy.
Next, the AI Agent parses the user’s request and creates an initial execution plan based on asset size, market conditions, and potential risk parameters. The system analyzes the relationship between the user’s goal and current market conditions in order to find the most suitable execution strategy.
Once the analysis is complete, the intent is broadcast to the Solver network. Multiple Solvers then try to find the optimal execution path at the same time, assessing liquidity sources, transaction costs, and expected slippage.
After verification, the best solution moves into the on-chain execution stage. When the trade is completed, the result is returned to the user interface, and the relevant asset status and portfolio allocation are updated. From the user’s perspective, the entire process appears as a simple goal input, while the underlying execution is handled automatically by the system.
A Solver can be understood as an intelligent execution engine within the Velvet ecosystem. Its core responsibility is to identify the execution path that best matches the user’s goal among many possible options.
To do this, the Solver analyzes information across several dimensions. First, the system evaluates the depth of different liquidity pools, because deeper liquidity usually results in lower price impact.
Second, the Solver compares the slippage costs that different routes may produce, helping improve price stability during execution. When a trade involves multiple chains or protocols, the system also calculates gas costs as well as the time and fees required for cross-chain bridging.
Execution success rate is another important factor. Rather than choosing a path with a better theoretical price but weaker reliability, the Solver often favors a route that can complete the trade more consistently. As a result, the best execution path is not necessarily the one with the lowest price alone, but the optimal outcome after considering total cost, efficiency, and success rate together.
Velvet does not rely on a single liquidity source. Instead, it connects multiple DeFi protocols and trading markets through an aggregation mechanism. The system can access decentralized exchanges, automated market maker pools, trading aggregators, and cross-chain liquidity networks at the same time, expanding the range of available liquidity.
This aggregation model allows Velvet to look for trading opportunities across a broader market environment. For ordinary trades, the system can automatically choose an execution path with better pricing and lower costs.
For large trades, Velvet can also reduce price impact by splitting trading routes or drawing from multiple liquidity sources, which improves overall execution efficiency. This capability is an important foundation for the automatic optimization made possible by Intent-Based Trading.
MEV(Maximal Extractable Value)is a common issue in on-chain trading environments. Attackers may take advantage of transaction ordering to carry out front-running, sandwich attacks, or other forms of value extraction.
Velvet’s Intent-Based Trading introduces a series of mechanisms designed to reduce these risks. Because users submit a final goal rather than a complete execution path, external observers have a harder time predicting transaction details in advance, which reduces the chance of being targeted.
The competition among multiple Solvers also helps reduce the risk that a single participant controls the execution process. At the same time, the system prioritizes liquidity sources with stronger security and stability, while using path optimization to reduce transaction exposure.
Although Intent-Based Trading cannot eliminate MEV completely, it can help mitigate the impact of related attacks to some extent and improve the user’s execution experience.
| Comparison Dimension | Intent-Based Trading | Traditional Swap |
|---|---|---|
| User input method | Expresses the final goal | Specifies the exact operation |
| Route planning | Automatically handled by the system | Handled by the user |
| Liquidity search | Automatically aggregated | Manually selected by the user |
| AI support | Supports AI Agents | Usually not supported |
| Multi-protocol coordination | Automatically handled | Operated independently by the user |
| Barrier to use | Relatively low | Relatively high |
| Execution efficiency | Higher | Depends on user experience |
The two models represent different stages of the DeFi user experience. Traditional swaps place greater emphasis on user control over the execution process, while Intent-Based Trading focuses more on automation and intelligent execution.
Velvet’s Intent-Based Trading uses AI Agents, Solver networks, and liquidity aggregation to turn complex on-chain trades that traditionally required manual user operation into simple expressions of intent. Users only need to describe the result they want, and the system automatically finds the best execution path and completes the trade. As an important part of the DeFAI ecosystem, this intent-driven model is pushing decentralized finance from manual operation toward intelligent execution, while laying the foundation for future financial services driven by AI Agents.
Intent-Based Trading requires users to express their final goal, while traditional trading requires users to specify the execution steps and trading route themselves. The system handles route planning and execution, reducing operational complexity for users.
Velvet uses AI Agents to parse user requests and generate execution plans, but final trades are still completed through on-chain infrastructure and the Solver network. AI mainly plays the role of understanding intent and optimizing the execution process.
A Solver is Velvet’s execution optimization module, responsible for finding the best path to achieve the user’s goal. The system compares factors such as price, liquidity, gas costs, slippage, and execution success rate.
One of the design goals of Intent-Based Trading is to simplify cross-chain operations, so it can work with cross-chain liquidity networks and bridging infrastructure, reducing the need for users to manually handle cross-chain processes.
Intent-Based Trading cannot fully eliminate MEV, but by hiding the execution path, introducing competition among Solvers, and optimizing the way trades are executed, it can reduce the impact of some MEV attacks.





