Original | Odaily Planet Daily

Author | 0x Sekiro

As the cornerstone of the blockchain world, DeFi has always given us the impression of being serious and professional. Only Smart Money and senior players on the chain can obtain fruits in the cold but prosperous code forest through complex interactive processes and formula calculations. However, the emergence of AI has made the interaction artistic and will have a profound change on the paradigm of DeFi and even Web3.

As Onekey founder Wang Yishi said, the narrative of Web3 has shifted from cryptocurrency to AI. It is not only AIGC that promotes the free creation atmosphere of Web3. More importantly, the "Intent-centric" thinking brought by AI is in turn promoting the changes in DeFi.

Whether it is the Bots that integrate DeFi products into Telegram through API interfaces, platforms like Dune and Nansen that access AI services on their respective platforms, or Anoma, an ecosystem with intent recognition as its core, we can broadly regard it as the promotion of the Web3 ecosystem by intent.

The emphasis on intent has given DeFi new front-end requirements and provided a new battlefield. Under the premise that Web3 traffic has peaked, whoever can capture user intent more efficiently will have more possibilities to break through the circle and gain incremental growth.

Intent is not a new concept

In the context of Web3, the early 1inch’s multi-DEX aggregation compared to Uniswap falls into the category of intention interpretation. For example, in the early days of DeFi, if there was no front end, if users wanted to buy an ETH on DEX with the greatest utility, they would need to manually look up the exchange prices of different exchanges and split their transactions according to the depth of different pools.

But in fact, we don’t need to care about the transaction path of Uniswap on 1inch. The intervention of 1inch, a DEX aggregator, will split the transaction to multiple exchanges to obtain the best results.

This is because 1inch understands the most basic intentions of the general public and optimizes based on DEX functions.

Generally speaking, the smart voice assistants we use in our daily lives, such as Siri or Bixby, are the most intuitive embodiment of Intent in the traditional world. Although they are not fully intelligent yet, the ability to clarify schedules based on a voice input has become an example of Intent.

"Hey, Siri, remind me to close the BTC contract at 10 am tomorrow." In this behavior, Siri plays the role of an intention interpreter. By interpreting the user's words, it pulls the relevant mobile applications and inputs instructions on behalf of the user. Except that it cannot trade.

If we transfer this paradigm to the Web3 context, we will find that many current Web3 products are still not as smart as Siri. When it comes to the complex logic of on-chain transactions, Siri and the traditional world do not provide us with more path references.

The reason why Siri can efficiently interpret the user's Intent is that the precondition is the strong integration of the IOS port, TB-level data and billions of applications are constantly trained, which can give Siri a lot of room to play. This condition is not available in the Web3 industry before.

Intentions are complex. Misjudgments can occur even in the interaction of intentions between people. The misjudgment between machines and people is likely to be even greater. However, the popularity of AI and large language models has made it possible to accurately identify intentions, which is why intention recognition has been valued in Web3 recently.

Three Levels of Intention

Let AI understand your intentions, which will bring changes to your Web3 transaction behavior.

In Paradigm’s article about AI, “Intent-Based Architectures and Their Risks”, there are different descriptions of trading methods, which can be expanded into three levels here. We can also have an intuitive understanding by classifying the current DeFi projects in the industry.

The first level, self-execution of trading strategies:

That is, the user manually completes the entire transaction process. If you want to quickly buy some LINK with USDC in Metamask, you currently have two options: the first method is to manually open the wallet, select the currency, add Gas, and click Swap.

This level is basically represented by most products before AI became widely popular.

The second level, the intention assistance of trading strategy:

This is the current mainstream idea in the industry. Based on limited intent assistance, users can more easily obtain the information or transaction results they need.

It is divided into three categories: one is the trading robot that accesses different DEX interfaces and is integrated into Telegram. The representatives of this category are currently mainly UniBot, Banana Gun and Maestro. They generally provide users with basic functions such as token sniping, copy trading, and limit order trading. In essence, it is based on the logic of conventional DEX to provide users with a more convenient DeFi interactive experience.

UniBot Operation Interface

The second category is based on the original functions of the project, expanding AI, adding or further optimizing products for intent interpretation. Representatives of this category include:

Uniswap X: Uniswap X is a permissionless protocol that uses the Dutch auction method. It focuses on solving the problem of dispersed liquidity by focusing on intention, helping users provide the best asset prices, and bringing gas-free transactions to users by wrapping the gas fee in the transaction price and letting the filler (the actual trader and liquidity provider of the user's transaction) bear it.

How Uniswap X works

Essential: Essential is an intent-based infrastructure that aims to capture user intent by building a unified processor network through a modular intent layer, thereby reducing user transaction costs.

Cowswap: CowSwap is a DEX developed by the Cow Protocol team. It adopts a batch auction price discovery mechanism, that is, processors compete with each other to provide users with the best transaction solutions to reduce users' transaction costs.

Soul wallet: It is a smart and friendly wallet that uses account abstraction technology to improve the flexibility of gas fee payment and sponsorship. The most commendable feature is the social recovery function, which means that there is no need for mnemonics. By signing a wallet recovery agreement with trusted members related to the account, the identity of the visitor can be determined as long as it is approved by the majority. In addition, Soul Wallet also provides two-factor verification, that is, users can use another wallet to approve transactions.

Similar to Soul Wallet are Argent Wallet and Unipass, which uses email to recover wallets.

Anoma: Anoma is a multi-purpose intent solution that also achieves the user's intention to seek lower transaction costs by allowing processors to compete with each other.

The capture of intent by the above two types of products is somewhat similar to the battle between WeChat and Alipay in the payment field. Although Alipay has sufficient first-mover advantage, it has been far ahead in the payment field in the early stage.

But WeChat, which entered the market later, has a staggering 1.34 billion monthly active users based on its social platform. Such a large number of users provides fertile soil for WeChat Pay. As a result, WeChat Pay’s 87.9% penetration rate surpassed Alipay’s 85.3% in 2022.

Changes in the market share of WeChat Pay and Alipay. Image from http://www.199it.com/archives/1550306.html

In Web3, bots led by UniBot, relying on Telegram's huge data of 700 million monthly active users, are challenging classic DEXs by building bots with different functions.

Classic DEXs, represented by Uniswap, are also rebuilding their intended moats in their own way.

The two have different approaches. Bots emphasize convenience. By aggregating different DEXs and integrating them into Telegram, the complexity of DeFi transactions is simplified to attract more users. However, due to the existence of Bot transaction fees (UniBot 1%, Banana Gun 0.5%), users will not feel a significant reduction in costs. Uniswap X will first consider the transaction costs of users.

In an article by ACCapital titled "Intention: The Starting Point of Web3 Interactive Intelligence", Anoma is considered a top-down iteration and Bot is considered a bottom-up revolution. This view also explains the current phenomenon.

But in the end they all move in the same direction, and even the Banana Gun block bribery system and the processor competition scheme have similarities.

That is, with the help of AI, the gap between products and users will be narrowed or even eliminated. This brings us to the third level described in the Paradigm article:

The third level: AI automatic execution of trading strategies:

Users no longer need to describe instructions in a concrete way, but give an abstract answer: I want to buy some LINK quickly, and the system will automatically identify the user's needs and. This level is the goal that no projects have emerged in the current Web3 industry, but everyone is working hard for.

Perhaps with the arrival of the third level, you can complete complex DeFi operations by simply saying a few words while eating breakfast.

Currently, some technology providers are exploring new paths for human-computer interaction based on Web3. For example, Hummingbot, 3 Commas, HaasOnline, Gekko, etc. are dedicated to assisting users in trading. These products are committed to helping users automatically execute trading strategies and support multiple DEX platforms currently on the market.

Or Nuo Network and Dharma Protocol, which have intelligent credit rating functions and may solve the problem of cryptocurrency credit lending.

In addition, there are Kaito AI, Messari, Nansen AI, etc. that assist users in decision-making and provide users with investment ideas.

The Future of Intent+Web3

Every iteration of user experience on the Internet is actually an improvement in the way the backend code is encapsulated in the frontend. As mentioned above, 1inch's DEX aggregation function allows users to automatically split your transactions into various liquidity pools with just a click, without having to worry about the execution logic behind the product.

Then, as the product captures intent to a higher degree, the corresponding front-end performance will be smarter. For example, Cow Protocol's Cow Hooks aggregates a variety of complex actions such as transactions, cross-chain, staking, deposits, etc. As long as the user is willing, the token's Approve, Swap and cross-chain functions can be realized with one click.

In addition, ChatGPT’s conversational interaction method has been widely used in many Web3 projects, although this type of conversation has not yet involved specific transaction behaviors. We can make a prospect that when the project’s intention capture is comprehensive enough and the front end is smart enough, it will not be difficult to see transaction behaviors based on conversations or even language patterns in the future. Perhaps in the future, dialog boxes with transaction functions will become the standard configuration of every DeFi project.

In this way, serious DeFi will no longer become aloof. Ordinary users, even novice users, who want to participate in DeFi products with complex rules such as Curve and GMX, no longer need to be familiar with cumbersome interaction processes or calculate complex formulas. They only need to state their needs to complete the transaction.

To a certain extent, this has also become an important path to break the circle in the Web3 field.