DeFAI and Intent: the future of decentralized finance (DeFi)?

Context: The Problem with DeFi

Decentralized Finance (DeFi) has opened up new possibilities by allowing users to access financial services without intermediaries. However, its potential adoption by the general public remains hindered by numerous constraints: a particularly complex user experience, inherent risks associated with on-chain infrastructures, and many others.

Interacting with DeFi protocols requires a certain level of technical knowledge: understanding how a wallet works, how smart contracts function, executing transactions on a blockchain, not to mention bridges to transition between ecosystems. Unlike traditional financial services where everything is largely simplified and intermediated, even the simplest operation in DeFi requires a series of manual transactions, each involving fees and a risk of error.

Moreover, a user looking to optimize their DeFi activity must not only select the right protocol but also follow multiple technical steps: approving a contract, depositing funds, managing impermanent loss in a liquidity pool, monitoring market trends, and repositioning their assets accordingly. This complexity represents a real barrier to adoption, even for experienced users.

Beyond operational complexity, DeFi also involves systemic risks that are poorly understood by users. Among them:

• MEV (Maximal Extractable Value): Bots and validators exploit user transactions by prioritizing or repositioning them in blocks to extract profit. This leads to hidden losses for users, particularly through front-running and sandwich attacks, where orders are manipulated to maximize gains for malicious actors.
• Transaction inefficiency: The optimal transaction route is rarely chosen for the user. DEX aggregators like 1inch or CoW Swap (although the latter is beginning to incorporate some of the mechanisms discussed in this analysis) improve the situation, but transactions remain limited to precise instructions, requiring direct user involvement at every step.
• Unpredictable transaction fees: Blockchain congestion leads to fee volatility, forcing users to constantly monitor costs before executing a transaction. On Ethereum, fees can skyrocket during high activity periods, making some operations unprofitable.

In this complex and risky environment, many protocols have made it a priority to rethink the DeFi infrastructure to make it more accessible and secure. Intents and DeFAI are emerging as solutions aimed at simplifying and automating on-chain interactions, reducing the technical decisions users must make while optimizing transaction execution.

→ Read our article on AI Agents:

AI agents: the next crypto trend?

In this analysis, we explore a narrative in trend: AI agents, through their origins, current applications and the flagship projects shaping this trend. What challenges and opportunities accompany their rise? What might an ecosystem powered by these autonomous entities look like? Let's dive into this rapidly expanding universe.

---

What Are Intents?

Definition

Intents represent a new approach to designing on-chain interactions. Instead of requiring users to execute a sequence of transactions to achieve a desired outcome, the concept of intent is based on a declaration of intention: the user defines a final objective, and the system finds the best way to achieve it.

An intent can be seen as a declarative instruction, where the user specifies what they want to obtain (e.g., buying ETH at a certain price and staking it for the best yield) without necessarily dictating the exact steps to get there. Unlike a traditional transaction, which follows a strict imperative order, an intent allows for multiple execution paths, as long as the results meet predefined constraints.

These declarations of intent are processed by decentralized networks (which we will discuss later) that employ solvers. The goal of these solvers is to find the most efficient path to accomplish the task requested by the user.

In other words, instead of manually executing multiple transactions to buy ETH and stake it, the user can simply express an intent such as: "I want to buy 1 ETH, paying a maximum of 2500 USDC, and stake it with a minimum yield of 4%," leaving the network to determine the best execution possible.

This paradigm shift introduces an abstraction that enhances user experience while enabling automated optimizations in transaction execution.

How It Works

The implementation of intents relies on an infrastructure where multiple actors interact to transform a declaration of intent into an executed on-chain transaction. It begins with the user interface generating the intent and sending it to network observer nodes.

These nodes then distribute the intent to solvers—entities responsible for determining the best possible execution. Solvers can be market makers, aggregation protocols, or actors leveraging specialized algorithms. They analyze various parameters such as available liquidity on DEXs, transaction costs, and optimization opportunities to propose the most efficient execution route.

Once the intent is resolved, the solver submits a single transaction that aligns with the user’s constraints. This transaction may involve multiple actions in one interaction, such as swapping an asset and staking it in a single step. The user only needs to sign this single transaction, avoiding a series of manual operations.

One of the major advantages of intents is their flexibility and efficiency. They allow for optimized transaction routing, lower gas costs, and mitigation of MEV-related attacks. This approach is particularly beneficial for advanced use cases such as multi-DEX trading, automated yield farming strategies, or cross-chain transactions.

However, this architecture introduces new dependencies, particularly concerning solvers. It is crucial to implement economic security measures to incentivize solvers to behave correctly. This could include slashing mechanisms or observers that verify whether the outcome of an intent truly matches the user’s request. Additionally, safeguards must be in place to prevent any single solver from dominating the network, as an efficient solver could gain continuous prominence and centralize the system.

→ To go further, our analysis of ai16z, the mine of AI agents:

A closer look at Ai16z (AI16Z): the mine of AI agents

If launchpads are often considered the picks and shovels necessary for creating AI agents, they are not indispensable for their emergence. However, there is one project around which the entire AI Agent narrative is being built—a project that could be likened to a mine where the essence of AI agents resides: ai16z.

---

What Is DeFAI?

Definition

DeFAI (sometimes referred to as "AgentFi") is a term derived from the combination of DeFi and AI, referring to the integration of artificial intelligence (particularly AI agents) into decentralized finance to automate and optimize on-chain interactions. Simply put, DeFAI is based on the idea that AI agents can autonomously execute DeFi strategies based on user preferences.

Just as ChatGPT has drastically revolutionized how we use the internet—allowing an AI to understand user intentions and aggregate information across multiple websites—AI agents can do the same with smart contracts on the blockchain.

The core principle of DeFAI is similar: instead of requiring a user to navigate through multiple DeFi protocols, they interact with an AI agent that acts on their behalf. These AI agents utilize machine learning models fine-tuned for financial applications, enabling them to understand user intentions and execute the necessary actions by interacting with smart contracts.

DeFAI has two main objectives: reduce DeFi complexity and enhance investment strategy efficiency on-chain. This approach can be applied to multiple areas, including portfolio management, transaction fee optimization, arbitrage, and liquidation protection.

How It Works

DeFAI architecture relies on AI agents that interact directly with smart contracts and DeFi protocols. These agents fulfill three key roles: data analysis, decision-making, and transaction execution.

Everything begins with data collection and analysis. The AI agent continuously gathers on-chain and off-chain information, including asset prices, available liquidity, market volatility, and transaction fees. Based on these data points, the agent evaluates different possible strategies and selects the one that maximizes returns while minimizing risks.

Once the strategy is defined, the agent executes transactions directly via smart contracts. Unlike traditional solutions that require manual interactions, DeFAI enables seamless, continuous execution without human intervention. The automation possibilities are endless: "Sell a token when the top 10 holders start selling," for example.

A key aspect of DeFAI is personalization and adaptability. Just as ChatGPT adapts to its user, a DeFAI agent could adjust based on an investor’s risk profile and historical activity.

Unlike intent-based networks, AI agents do not require economic security mechanisms at the network level. Simple solutions exist, such as integrating AI agents within Trusted Execution Environments (TEE) to prevent external influence.

Ultimately, DeFAI adds a truly intelligent layer to the blockchain (in contrast to the misleading term "smart contracts"). This approach introduces a new way to interact with DeFi: more intuitive, more efficient, and more automated.

---

Convergence Between DeFAI and Intents

A Clear Complementarity

The concept of intents emerged in 2021 and has gained traction in recent months. However, its rapid acceleration coincides with the rise of DeFAI, and this is no coincidence—both approaches share a common goal: simplifying and automating the DeFi user experience.

As we previously explained, intents rely on a network of solvers capable of handling user requests. Currently, these are algorithms designed to optimize transactions based on a series of predefined parameters. However, they are still limited in their ability to interpret complex intentions and dynamically adapt to market conditions.

This is precisely where DeFAI comes into play. By integrating AI agents into solvers, it becomes possible to introduce a more advanced understanding of user needs and real-time adaptability. An AI agent can interpret an intent with greater nuance than a traditional solver, taking into account not just transactional parameters but also factors like the user’s investment profile, trading history, and broader market trends.

The addition of intelligence to solvers is not just an ergonomic improvement—it represents a structural evolution of DeFi. With the integration of DeFAI into intent-based systems, we are moving towards programmable finance, where the user no longer defines individual transactions but rather entire strategies, dynamically adjusted in real time by autonomous AI.

A Functional Example

The implementation of DeFAI, particularly when integrated into existing intent networks, can be divided into three key layers:

• The Abstraction Layer: The application where the end user connects their wallet and formulates a clear intent request (e.g., "Swap 500 USDC for ETH and stake it").
• The Intelligence Layer: AI models designed to translate the user’s intent into a series of actions. These models are fine-tuned LLMs specializing in different DeFi sectors.
• The Execution Layer: The network of solver agents responsible for executing transactions on-chain according to the intelligence layer’s instructions.

→ To find out more about Nous Research, an exciting project :

Nous Research and Psyche, the open-source, decentralized AI revolution

Nous Research, a collective of artificial intelligence researchers and developers, has just announced the launch of Psyche on Solana. Here's an overview of this new decentralized platform and other Nous Research products.

Key Projects to Watch

The convergence between Intents and DeFAI is already underway, driven by several projects exploring different facets of this synergy. Here are some of the leading protocols shaping this new approach in the DeFi ecosystem.

1) Anoma: A Full-Intent Interoperable Architecture

Anoma is one of the first protocols to implement a full-intents infrastructure, aiming to make DeFi interactions more seamless and interoperable. Unlike traditional blockchains where users submit transactions, Anoma enables the submission of generalized intents, which can be optimally resolved by a network of solvers.

One of Anoma’s major strengths is its ability to operate across multiple ecosystems. Intents and applications built with Anoma can be executed on Ethereum, EVM and non-EVM rollups, EigenLayer AVS, Cosmos, and Solana, making it particularly useful for cross-chain transactions.

Anoma operates through a decentralized architecture involving multiple actors:

• Users submit intents via a simplified interface.
• Solvers analyze execution options and determine the best route.
• Validators ensure transactions comply with the conditions defined by the intents.

2) Aperture Finance: AI-Integrated Solvers

Aperture Finance takes the intent model a step further by incorporating an AI layer to optimize transaction execution. Its goal is to enhance user experience by enabling solvers to factor in advanced parameters, such as:

• Fluctuating transaction fees across different blockchains.
• Liquidity availability across multiple platforms.
• Risk management and optimal execution based on market conditions.

Aperture’s approach involves intelligent solvers that dynamically adapt to the DeFi environment. Instead of merely executing transactions, an AI-enhanced solver can make more sophisticated decisions, such as:

• Distributing funds across multiple yield strategies.
• Deploying liquidation protection mechanisms in volatile markets.
• Adjusting transactions based on the user’s risk profile.

3) Fetch.AI: Autonomous AI Agents for DeFi

Fetch.AI is one of the most advanced projects in autonomous AI agents for decentralized finance. Unlike traditional solvers, Fetch.AI introduces a framework of AI agents that can independently make decisions and interact directly with blockchains.

These AI agents monitor market conditions and execute transactions in real time. Their most advanced capabilities include:

• Automated arbitrage across multiple protocols to maximize yield.
• Proactive risk management, dynamically adjusting positions based on market fluctuations.
• On-chain trend monitoring, enabling execution before major market shifts occur.

4) HiveAI and Griffain: AI in the Solana Ecosystem

Solana has become a hub for the development of AI-driven DeFi automation. Two standout projects in this sector are HiveAI and Griffain.

• HiveAI is creating an ecosystem where AI agents collaborate to execute complex financial strategies. For instance, an agent might track real-time memecoin trends on Solana and automatically adjust a portfolio to maximize gains.
• Griffain focuses on automated asset management, allowing users to delegate full portfolio management to an AI agent that optimizes allocations based on market conditions.

Towards Mass DeFi Adoption?

Projects like Anoma, Aperture Finance, Fetch.AI, HiveAI, and Griffain—as well as others like Anon, Orbit, Neur.sh, and Hiero Terminal—indicate that the future of decentralized finance may rely on intelligent automation and deeper abstraction.

While intents simplify the expression of user needs, DeFAI introduces autonomous intelligence capable of executing strategies efficiently and proactively. If these technologies continue to evolve, they could significantly lower DeFi’s entry barrier, making decentralized financial services as accessible as traditional Web2 applications.

However, several challenges remain:

• Governing and decentralizing AI solvers to prevent excessive power concentration.
• Ensuring security and reliability in AI-driven decision-making to minimize financial risks for users.
• Seamlessly integrating these innovations into existing DeFi infrastructure to ensure broad adoption.

Despite these obstacles, AI-driven automation and optimization are rapidly reshaping the DeFi experience, laying the groundwork for mass adoption and a more accessible, efficient, and scalable decentralized financial ecosystem.

---

Conclusion

The future of DeFi depends on its ability to simplify and automate the user experience. Whether through AI agents or decentralized networks, intent-based systems represent a necessary step toward the democratization of DeFi. Rather than being competing solutions, AI agents and intents can actually converge, merging transactional optimization with autonomous intelligence.

However, challenges remain. While AI agents do not pose the same risks as malicious solvers in an intent network, they can still make errors and cause financial losses for users.

That said, AI is already making its way into blockchain ecosystems. Given that AI has transformed nearly every other industry, there’s no reason why it won’t eventually revolutionize blockchain as well.

With the rapid advancement of models from Google, OpenAI, and other tech leaders, it is difficult to predict exactly what these models will be capable of in the future. However, one thing is certain: they will significantly simplify blockchain’s user experience, which remains one of the biggest barriers to adoption.