
The Intelligent Evolution of DeFi: The Evolutionary Path from Automation to AgentFi
TechFlow Selected TechFlow Selected

The Intelligent Evolution of DeFi: The Evolutionary Path from Automation to AgentFi
This article will focus on the convergent evolution of DeFi and AI, outline their progression from automation to intelligence, and analyze the infrastructure, application scenarios, and key challenges of strategy-execution agents.
Author: 0xjacobzhao and ChatGPT 4o
Special thanks to Lex Sokolin (Generative Ventures), Stepan Gershuni (cyber.fund), and Advait Jayant (Aivos Labs) for their valuable feedback on this article. Input was also gathered from project teams including Giza, Theoriq, Olas, HeyElsa, Almanak, and Brahma.fi during the writing process. This article strives for objectivity and accuracy; however, as some perspectives involve subjective judgment, biases may exist. Readers are encouraged to critically engage with the content.
In today's crypto industry, stablecoin payments and DeFi applications are among the few use cases already proven to have real demand and long-term value. Meanwhile, AI Agents are rapidly emerging as the primary user-facing interface in the AI industry, serving as a key intermediary layer connecting AI capabilities with user needs.
In the convergence of Crypto and AI—particularly where AI enhances crypto applications—current exploration centers around three typical scenarios:
-
Conversational Agents: Primarily focused on chat, companionship, and assistant functions. Though most remain wrappers around general large models, their low development barrier, natural interaction, and token incentives have made them the earliest form to gain market attention.
-
Information Integration Agents: Focused on intelligent aggregation of online and on-chain data. Projects like Kaito and AIXBT have succeeded in off-chain information search, while on-chain data integration remains exploratory with no clear leaders yet.
-
Strategy Execution Agents: Centered on stablecoin payments and DeFi strategy execution, evolving into Agent Payment and DeFAI. These agents deeply embed on-chain trading and asset management logic, potentially breaking past hype cycles to become intelligent execution infrastructure with financial efficiency and sustainable returns.
This article focuses on the evolution of DeFi and AI integration, outlining its progression from automation to intelligence, analyzing the infrastructure, application scope, and key challenges of strategy-execution Agents.
Three Stages of DeFi Intelligence: Automation, Copilot, and the Leap to AgentFi
In the evolution of DeFi intelligence, system capabilities can be divided into three stages: Automation (automated tools), Intent-Centric Copilot (intent-driven assistants), and AgentFi (on-chain agents).

-
Automation resembles a rule trigger: Executes fixed tasks based on predefined conditions—such as arbitrage, rebalancing, take-profit, or stop-loss—but cannot generate strategies or operate independently.
-
Copilot introduces intent recognition and semantic parsing. Users input via natural language, the system interprets and suggests execution paths, but final confirmation is still required, leaving the execution loop open.
-
AgentFi represents a complete intelligent loop of “perception → reasoning/strategy generation → on-chain execution → evolution,” embodying autonomous on-chain agents capable of self-sustained operation and continuous evolution.

To determine whether a project truly qualifies as AgentFi, it must meet at least three of the following five core criteria:
-
Autonomous perception of on-chain states/market signals (not static input, but real-time monitoring)
-
Ability to generate and compose strategies (not predefined, but capable of self-formulating action plans based on context)
-
Capable of autonomous on-chain execution (no user interaction required, able to perform complex operations such as swap/lend/stake)
-
Persistent state and evolutionary capability (the agent has a lifecycle, can run long-term, and self-adjust based on feedback)
-
Agent-native architecture (e.g., dedicated Agent SDK,托管 execution environment, Agent middleware)
In other words, automated trading ≠ Copilot, and certainly ≠ AgentFi: Automated trading is merely a "rule trigger," while Copilot, though capable of understanding user intent and suggesting actions, still depends on human involvement. True AgentFi refers to "intelligent agents with perception, reasoning, and autonomous on-chain execution," capable of completing closed-loop strategies and continuous evolution without manual intervention.
Analysis of DeFi Scenario Suitability for Intelligence:
Within the DeFi (decentralized finance) ecosystem, core applications can broadly be categorized into asset circulation/exchange and yield-generating financial services. We believe these two categories differ significantly in their suitability for intelligent adaptation:
1. Asset Circulation and Exchange Scenarios
These scenarios are primarily atomic interactions—including swaps, cross-chain bridges, fiat on/off ramps—and are characterized by "intent-driven + single atomic interaction." Transactions do not involve yield strategies, state maintenance, or evolutionary logic, making them better suited for lightweight execution via Intent-Centric Copilot rather than AgentFi.
Due to low engineering barriers and simple interactions, most current DeFAI projects fall into this category, which does not constitute a closed-loop AgentFi agent. However, certain advanced complex swap strategies (e.g., cross-asset arbitrage, perpetual hedging LPs, leveraged rebalancing) could benefit from AI Agent integration, though this remains in early exploration.

2. Yield-Generating Financial Scenarios
Yield-generating scenarios feature clear return objectives, complex strategy combinations, and dynamic state management needs—naturally aligning with AgentFi’s “strategy loop + autonomous execution” model. Key characteristics include:
-
Quantifiable return goals (APR/APY), enabling agents to establish optimization functions;
-
Vast strategy combination space across multiple assets, timeframes, platforms, and interaction flows;
-
Frequent, real-time operational adjustments suitable for execution and maintenance by on-chain agents.

Due to factors such as yield duration, volatility frequency, on-chain data complexity, cross-protocol integration difficulty, and compliance constraints, different yield scenarios vary significantly in their suitability and feasibility for AgentFi implementation. Priority recommendations are as follows:
High-Priority Implementation Directions:
-
Lending/Borrowing: Interest rate fluctuations are easy to track with standardized execution logic, ideal for lightweight agents.
-
Yield Farming: Pools change frequently, offer vast strategy combinations and high return volatility—AgentFi can significantly optimize annualized returns and interaction efficiency, though engineering implementation poses challenges.
Mid-to-Long-Term Exploratory Directions:
-
Pendle Yield Trading: Clear time dimensions and yield curves make it ideal for Agent-managed maturity rollover and inter-pool arbitrage.
-
Funding Rate Arbitrage: Theoretically attractive returns, but requires solving cross-market execution and off-chain interaction challenges—high engineering complexity.
-
LRT Dynamic Portfolio Structures: Static staking is unsuitable, but automatic adjustment of strategies combining LRT + LP + Lending is possible.
-
RWA Multi-Asset Portfolio Management: Hard to deploy short-term, but Agents can assist in portfolio optimization and maturity-based strategies.
Introduction to Intelligent DeFi Projects:
1. Automation Infrastructure: Rule Triggers and Conditional Execution
Gelato was one of the earliest DeFi automation infrastructures, previously supporting conditional task execution for protocols like Aave and Reflexer, but has since pivoted to Rollup-as-a-Service. Today, the main battleground for on-chain automation has shifted to DeFi asset management platforms (DeFi Saver, Instadapp). These platforms integrate standardized auto-execution modules such as limit orders, liquidation protection, automatic rebalancing, DCA, and grid strategies. Additionally, some more complex DeFi automation tool platforms have emerged:
Mimic.fi (https://www.mimic.fi/)
Mimic.fi is an on-chain automation platform serving DeFi developers and projects, enabling programmable automated tasks on chains like Arbitrum, Base, and Optimism. Its core uses "if-then" rule triggers for cross-protocol automated execution, structured in three layers: Planning (task and trigger definition), Execution (intent broadcasting and execution bidding), and Security (triple verification and safety controls). Currently accessed via SDK, the product is still in early deployment.
AFI Protocol (https://www.afiprotocol.ai/)
AFI Protocol is an algorithm-driven Agent execution network supporting 7×24 non-custodial automated operations, aiming to solve issues of fragmented execution, high strategy barriers, and risk response in DeFi. Designed for institutions and advanced users, it offers orchestratable strategies, permission management, and SDK tools, and has launched a yield-bearing stablecoin afiUSD as its native asset. Currently in Sonic Labs’ private test phase, not yet publicly launched or available to retail users.
2. Intent-Centric Copilot: Intent Expression and Execution Suggestions
The DeFAI concept, popular toward the end of 2024, after excluding meme-token-driven speculation, mostly consisted of Intent-Centric Copilot-type projects—using natural language to express user intent, with systems responding with trade suggestions or basic on-chain operations. Core capabilities remain at the stage of “intent recognition + Copilot-style assistance,” lacking full strategy closure and continuous optimization. Many products face shortcomings in semantic understanding, cross-protocol calls, and feedback responsiveness, resulting in generally poor user experience and limited functionality.
HeyElsa (https://app.heyelsa.ai/)
HeyElsa is a Web3-focused AI Copilot that empowers users via natural language to perform various on-chain actions including trading, cross-chain bridging, NFT purchases, stop-loss setup, and Zora token creation. As a multi-functional conversational crypto assistant, it caters to users ranging from beginners to advanced traders (including highly active degen communities), currently supporting real-time interactions across over 10 major blockchains. The platform now sees daily trading volume reaching $1 million, with daily active users between 3,000 and 5,000. It has integrated yield optimization and automated intent execution modules, laying the foundational framework for AgentFi applications.
Bankr (https://bankr.bot/)
Bankr is an intent-based trading assistant integrating AI, DeFi, and social contexts. Users can issue commands via X platform or a dedicated terminal using natural language to execute swaps, limit orders, cross-chain bridges, token launches, and NFT minting, supporting Base, Solana, Polygon, and Ethereum. Bankr builds a full Intent → compilation → execution pipeline, emphasizing seamless, minimalist transaction experiences within social environments, and activates its ecosystem through token incentives and revenue sharing.
Griffain (https://griffain.com/)
Griffain is a multi-functional AI Agent platform deployed on Solana, allowing users to interact with Griffain Copilot via natural language to perform asset queries, swaps, NFT trades, and LP management. The platform includes multiple built-in Agent modules and encourages community participation in creating and sharing Agents. Built on Anchor Framework and components like Jupiter and Tensor, it emphasizes mobile adaptability and frontend composability. Currently supports over 10 core Agent modules, demonstrating strong execution capability and ecosystem integration.
Symphony (https://www.symphony.io/)
Symphony is on-chain execution infrastructure for AI Agents, building a full-stack system covering intent modeling, intelligent path discovery, RFQ execution, and account abstraction, aiming to become a core module in DeFi’s smart execution layer. The platform has launched a conversational assistant Sympson with market data lookup and strategy suggestion features, though on-chain execution is not yet live. Symphony provides essential components for AgentFi and will support multi-Agent collaboration and cross-chain operations in the future.
Hey Anon (https://heyanon.ai/)
HeyAnon is a DeFAI platform combining intent interaction, on-chain execution, and intelligence analysis, supporting multi-chain deployment (Ethereum, Base, Solana, etc.) and cross-chain bridging (LayerZero, deBridge). Users can perform swaps, lending, staking, and receive on-chain sentiment and market dynamics analysis via natural language. Despite high visibility due to founder Sesta, the project remains in the Copilot stage, with core strategy and execution intelligence not fully realized—its long-term trajectory remains to be seen.

The above scoring system is based on current product availability, user experience, and feasibility of public roadmaps, and involves subjectivity. Note that this assessment does not include code security audits and should not be taken as investment advice.
3. AgentFi Agents: Strategy Closure and Autonomous Execution
We believe AgentFi represents a higher-level evolution beyond Intent Copilot in the journey toward DeFi intelligence. Agents possess independent yield strategies and on-chain autonomous execution, significantly improving strategy efficiency and capital utilization. In 2025, we are pleased to see increasing AgentFi projects launching or planning products, primarily focusing on lending and yield farming. Representative projects include Giza ARMA, Theoriq AlphaSwarm, Almanak, Brahma, Olas series, and others.
Giza ARMA (https://arma.xyz/)
ARMA is Giza’s intelligent agent product, designed specifically for cross-protocol stablecoin yield optimization. Deployed on Base, it supports major lending protocols including Aave, Morpho, Compound, and Moonwell, with core capabilities in cross-protocol rebalancing, auto-compounding, and smart currency swapping. ARMA’s strategy system continuously monitors stablecoin APR, transaction costs, and yield differences, automatically adjusting capital allocation—achieving returns demonstrably higher than static holdings. Its architecture comprises smart accounts, session keys, core agent logic, protocol integrations, risk management, and accounting modules, ensuring secure and efficient non-custodial automated execution.
ARMA is now fully live and under continuous iteration. With its modular design, robust security mechanisms, and strong early performance data, ARMA stands as one of the most practical Agent products in DeFi yield management—one of the few AgentFi projects combining deep conceptual grounding with real-world utility.
Reference report: "A New Paradigm for Stablecoin Yield: From AgentFi to XenoFi" link: https://x.com/0xjacobzhao/status/1925226999699964158
Theoriq (https://www.theoriq.ai/)
Theoriq Alpha Protocol is a multi-agent collaboration protocol focused on DeFi scenarios. Its flagship product, Alpha Swarm, targets liquidity management, aiming to build a full-chain automated loop of “perception → decision → execution.” Composed of three types of Agents—Portal (on-chain signal sensing), Knowledge (data analysis and strategy selection), and LP Assistant (strategy execution)—it enables dynamic asset allocation and yield optimization without human intervention. The underlying Alpha Protocol provides Agent registration, communication, parameter configuration, and developer tools, forming the operational foundation of the entire Swarm system—viewed as a “smart agent operating system” for DeFi. Through AlphaStudio, users can browse, invoke, and combine various Agents to build modular, scalable automated trading networks.
As one of the first projects on Kaito Capital Launchpad, Theoriq recently raised $84 million from the community and is approaching TGE. Theoriq has recently launched the AlphaSwarm Community Beta testnet, with the mainnet version imminent.
Reference report: "Theoriq Research Report: The Evolution of AgentFi in Yield Farming" link: https://x.com/0xjacobzhao/status/1948545449016918511
Almanak (https://almanak.co/)
Almanak is an intelligent Agent platform for DeFi strategy automation, combining non-custodial security architecture with a Python-based strategy engine to help traders and developers deploy persistent on-chain strategies.
The core platform consists of Deployment (execution components), Strategy (logic), Wallet (Safe+Zodiac security modules), and Vault (strategy assetization), supporting yield optimization, cross-protocol interactions, liquidity provision, and automated trading. Compared to traditional DeFi tools, Almanak emphasizes AI-enhanced market perception and risk management, already achieving 24/7 intelligent operation, with plans to introduce multi-agent and AI decision systems—aiming to build next-generation AgentFi infrastructure.
Almanak’s strategy system is a state-machine program built in Python, acting as each Agent’s “decision brain,” autonomously formulating and executing on-chain actions based on market data, wallet status, and user-defined conditions. The platform provides a complete Strategy Framework, supporting encapsulated modules (Action Bundles) for on-chain trading, lending, and liquidity provision—without requiring low-level contract coding—and ensures strategy privacy and operational security through encryption isolation, permission control, and monitoring. Users can write strategies via SDK, with future support for natural language strategy creation, enabling a smooth transition from complex logic to no-code experience.
The product has already launched a USDC lending vault on Ethereum mainnet, while more complex trading strategies are in testing and require whitelist access. Almanak is set to join cookie.fun’s cSNAPS campaign for community fundraising—an event worth watching.
Brahma (https://brahma.fi/)
Brahma positions itself as the “Intelligent Capital Orchestration Layer” for internet finance, aiming to abstract on-chain accounts, execution logic, and off-chain payment flows to help users and developers efficiently coordinate on-chain and real-world assets. Through Smart Accounts, continuously running on-chain Agents, and a Capital Orchestration Stack, Brahma delivers intelligent fund management without backend maintenance.
Currently launched representative Agents include:
-
Felix Agent: Automatically optimizes feUSD debt interest rates, prevents liquidation, and saves on interest;
-
Surge & Purge Agent: Tracks volatility and executes automatic trades;
-
Morpho Agent: Deploys and rebalances funds in Morpho vaults;
-
ConsoleKit Framework: Supports integration of any AI model, unifying strategy execution and asset management.
Olas (https://olas.network/)
Olas Network’s AgentFi products, the BabyDegen series—including Modius Agent and Optimus Agent—are already deployed on-chain, covering multi-chain ecosystems (Solana, Mode, Optimism, Base), and feature full on-chain interaction, strategy execution, and autonomous asset management.
-
BabyDegen is an AI trading agent on Solana, implementing automated buy/sell based on CoinGecko data and community strategy libraries. Currently integrated with Jupiter DEX and in Alpha testing.
-
Modius Agent targets the Mode network, focusing on USDC and ETH portfolio management. Integrated with Balancer, Sturdy, Velodrome, it enables 24/7 automatic strategy execution after user preference setup.
-
Optimus Agent supports Mode, Optimism, and Base, integrating more protocols like Uniswap and Velodrome, offering flexible multi-chain strategy combinations suitable for intermediate to advanced users building automated asset management systems.
Axal (https://www.getaxal.com/)
Axal’s core product, Autopilot Yield, offers a one-stop, non-custodial, verifiable yield management experience, integrating mainstream protocols such as Aave, Morpho, Kamino, Pendle, and Hyperliquid. Designed around on-chain strategy execution and risk control, it empowers ordinary users to easily access complex on-chain yield networks.
-
Conservative Strategy: Focuses on low-risk, mainstream stable yield scenarios, deploying capital primarily on battle-tested platforms like Aave and Morpho, yielding ~5–7% APY. Achieves steady growth through TVL monitoring, stop-loss mechanisms, and top-strategy filtering—ideal for users prioritizing capital safety and long-term returns.
-
Balanced Strategy: Offers moderate risk and higher return potential (10–20% APY), leveraging wrapped stablecoins (e.g., feUSD, USDxL), liquidity provision, and neutral arbitrage positions. More diverse and complex in composition, exposure is managed through Axal’s automated monitoring and dynamic adjustments.
-
Aggressive Strategy: Targets high-risk, high-return users, incorporating high-leverage LPs, cross-platform chaining, low-liquidity market making, and volatility capture—potentially exceeding 50% APY. Axal’s smart agents implement stop-loss, auto-exit, and redeployment logic at the strategy level, providing a last line of defense in high-risk environments.
Fungi.ag(https://fungi.ag/)
Fungi.ag is a fully automated AI Agent optimized for USDC yield, automatically reallocating funds across lending protocols like Aave, Morpho, Moonwell, and Fluid to achieve optimal capital allocation based on yield, fees, and risk. Users need only authorize a Session Key to enable non-custodial automated strategy execution. Currently supports Base chain, with plans to expand to Arbitrum and Optimism. Fungi also opens the Hypha custom strategy scripting interface, allowing community development of strategies like DCA and arbitrage, and fosters a co-built ecosystem via DAO and social platforms.
ZyFAI (https://www.zyf.ai/)
ZyFAI is a DeFi intelligent assistant platform deployed on Base and Sonic networks, combining on-chain interfaces with AI assistance to help users manage assets intelligently under different risk profiles. Its core strategies fall into three categories:
-
Safe Strategy: Designed for conservative users, focusing on audited, established protocols like Aave, Morpho, Compound, Moonwell, and Spark, emphasizing unilateral USDC deposits and stable yield opportunities, prioritizing asset security and long-term reliability.
-
Yieldor Strategy: For high-risk users, unlocked only after holding 20,000 ZFI tokens. Covers high-yield protocols including Pendle, YieldFi, Harvest Finance, Wasabi, supporting complex strategies like DEX LP, yield splitting, and leveraged vaults—with future expansion into Looping and Delta-neutral structured products.
-
Airdrop Strategy: A future strategy currently under development, aimed at maximizing airdrop incentives.

The above scoring system is based on current product availability, user experience, and feasibility of public roadmaps, and involves subjectivity. Note that this assessment does not include code security audits and should not be taken as investment advice.
Practical Paths and Advanced Visions for AgentFi
Undoubtedly, Lending and Yield Farming are the most valuable and immediately feasible business scenarios for AgentFi. These areas are already mature in the DeFi world and naturally suit Agent integration due to the following shared characteristics:
-
Vast strategy space with many optimization dimensions: Beyond chasing highest yields, lending allows strategies like interest rate arbitrage, leveraged looping, debt refinancing, and liquidation protection;
-
Yield farming offers rich strategy orchestration including APR tracking, LP rebalancing, compounding, and strategy combinations.
-
Highly dynamic, ideal for real-time agent perception and response: Interest rate changes, TVL fluctuations, reward structure updates, new pools, and new protocols all impact optimal strategy paths, requiring dynamic adjustments.
-
Execution window opportunity cost makes automation highly valuable: Funds not allocated to optimal pools drag down returns—requiring automatic migration.
Notably, lending Agents, due to stable data structures and relatively simple strategies, have high feasibility—projects like Giza’s Arma have already launched. In contrast, yield farming management demands real-time response to price volatility, changing volatility, and fee accumulation, placing extremely high demands on an Agent’s data perception, strategic judgment, and on-chain execution. LP Agents must not only accurately predict market states but also dynamically rebalance positions and redistribute yields on-chain—posing high engineering complexity, a challenge being tackled by projects like Theoriq.
Beyond lending and yield farming, here are longer-term exploratory directions based on AgentFi adaptability:
Pendle Yield Trading: Clear time dimensions and yield curves make it ideal for Agent-managed maturity rollover and inter-pool arbitrage
Pendle’s unique structure of “yield splitting + maturity mechanism + yield rights trading” provides natural strategy orchestration space for AgentFi. Assets are split into PT (Principal Token) and YT (Yield Token): PT represents redeemable principal at maturity, suitable for stable fixed-income allocation; YT represents yield rights, volatile and usable for speculation, farming, and arbitrage. Around these, users can build complex strategies such as fixed-income holdings, YT farming, maturity fund management, spread arbitrage, and portfolio hedging.
In practice, Pendle faces several user pain points urgently needing AgentFi solutions: High-yield pools are often short-term (1–3 months), requiring manual reconfiguration upon maturity; YT yields fluctuate widely across pools, making tracking and rotation costly; PT+YT combo strategies involve complex pricing and rebalancing. If AgentFi could automate the full workflow—from strategy identification and liquidity allocation to maturity rollover and redeployment—based on user return preferences and risk tolerance—it would dramatically improve capital efficiency and user experience.
Pendle’s “temporal, divisible, dynamic” triple characteristics perfectly match AgentFi’s strategy expression and execution pathways, especially in auto-compounding, implied yield arbitrage, and pool rotation—offering high-frequency, high-strategy complexity ideal for building “yield agent swarms” or Portfolio Agent systems. If combined with intent expression (e.g., “10% APY, withdrawable in 6 months”) and automated execution frameworks, Pendle could become one of the most representative AgentFi modules.
Funding Rate Arbitrage: Theoretically high returns, but requires solving cross-market, cross-chain interaction challenges—high engineering complexity
While the on-chain options sector has cooled due to pricing gaps, complex exercise, and poor composability, perpetual contracts remain one of the most active derivative scenes on-chain, offering entry points for AgentFi. Strategies around funding rate arbitrage, basis trading, and multi-platform hedging allow AgentFi to leverage intelligent perception, judgment, execution, and portfolio management.
Architecturally, AgentFi could embed four key modules: First, a data perception module capturing real-time on-chain and CEX funding rates, position costs, and market depth; second, an intelligent decision module dynamically determining entry/rebalancing based on arbitrage thresholds, leverage levels, and liquidation boundaries; third, an auto-execution module triggering position deployment or profit-taking once conditions are met; fourth, a portfolio management module enabling coordinated scheduling across chains, accounts, and strategies.
Real-world challenges include: First, current on-chain AgentFi focuses on smart contract interaction, lacking universal frameworks to directly connect to CEX APIs; second, high-frequency strategies demand extreme efficiency in execution, gas cost, and slippage control; third, complex arbitrage often requires multiple Agents working together, necessitating swarm-style collaboration.
Ethena’s funding rate arbitrage already relies on a highly automated execution system. While Ethena currently lacks AgentFi traits, if it were to further open its strategy modules, build a distributed Agent Swarm, and enable intent-driven goal expression, its system could naturally evolve into a full AgentFi infrastructure.
Staking and Restaking: Not naturally suited for AgentFi, but dynamic LRT portfolios offer some potential
Overall, traditional Staking and Restaking are not ideal AgentFi use cases, as single-chain staking involves simple operations, stable returns, singular decisions, and long unstaking wait times—unsupportive of the intelligent value emphasized by AgentFi.
However, in more complex staking constructs, AgentFi has limited applicability: (1) Focus on composable LST/LRT assets (e.g., stETH, rsETH), avoiding direct native ETH unstaking; (2) Build restaking + collateral + derivatives combo strategies, bypassing time lag from unstaking; (3) Deploy monitoring Agents for continuous optimization, dynamically assessing AVS risks, APR changes, and rebalancing positions.
Moreover, the restaking sector faces structural challenges: Rapid cooling in market enthusiasm, severe imbalance between supply (staked ETH) and demand (AVS security needs), and lack of real-world use cases for asset leasing. Leading projects like EigenLayer and Ether.fi have begun pivoting. Thus, Staking/Restaking may become modular strategy components in AgentFi rather than core application scenarios.
RWA Assets: U.S. Treasury protocols are not ideal, but multi-asset portfolio structures offer exploration value
Current mainstream RWA protocols typically use U.S. Treasuries (T-bills) as underlying assets, focusing on delivering stable, secure, compliant on-chain yield. However, from an AgentFi perspective, these products are unsuitable for intelligent agent integration due to stable asset nature (~4–5% APY with minimal spread, little room for optimization), low operation frequency (fixed lock-up and reinvestment cycles, unsuitable for frequent rotation or high-frequency compounding), and strong compliance constraints (KYC and regional restrictions). Additionally, non-interoperable asset structures across protocols limit Agent routing and liquidity aggregation.
Nonetheless, several potential directions could serve as mid-to-long-term AgentFi expansion paths:
1. Multi-Asset RWA Allocation Agent (RWA Multi-Asset Portfolio): As RWA products expand into real estate, credit bonds, and receivables, users may express intent to “allocate a basket of stable-yield assets with periodic rebalancing.” Allocation Agents could periodically adjust weights and redeploy maturing assets, building long-term yield stabilizers.
2. RWA and DeFi Integration (RWA-as-Collateral & custody reuse): Some protocols are exploring using tokenized T-bills as collateral in DeFi lending. Here, Agents could assist users in automatic deposits, rate comparisons, and collateral rebalancing, creating dual-income streams. If RWA assets achieve broad circulation on platforms like Pendle and Uniswap, Agents could track cross-platform premium/discount and implied yield changes, building auto-arbitrage and rolling deployment strategies. As markets mature, this could become a key breakthrough for AgentFi in the RWA space.
Swap Transaction Combos: Upgrading from Intent Infrastructure to AgentFi Strategy Engine
In today’s DeFi intelligence ecosystem, Swap transactions, enhanced by account abstraction and intent patterns, hide complex multi-chain DEX path choices, simplifying user input and lowering interaction barriers. However, these systems remain at the level of “atomic action automation,” lacking real-time environmental perception, responsive adaptation, or goal-oriented strategy execution—thus falling short of AgentFi’s intelligent agent criteria.
Under the AgentFi framework, Swap is no longer a single action but part of larger composite strategies. For example, when a user says “I want to configure stETH and USDC for maximum yield,” the Agent could automatically perform multiple swaps (USDC → ETH → stETH), restake, split Pendle PT/YT, deploy arbitrage strategies, and collect returns.
Further, Swap plays a critical role in three AgentFi scenarios:
-
Component of yield strategies: As a capital routing relay, Swap enables Agents to automatically execute asset allocation paths, improving strategy efficiency.
-
Cross-market arbitrage / delta-neutral strategies: By comparing prices across on-chain sources, Agents can dynamically adjust positions and build hedging portfolios.
-
Transaction risk defense: Upon detecting large trades, Agents can assess slippage, batch execute, and avoid potential MEV attacks.
Thus, a truly AgentFi-capable Swap Agent must possess dynamic strategy awareness, cross-protocol coordination, optimal fund routing, timing judgment, and risk prevention. Future Swap Agents should serve multi-strategy portfolios, dynamic position adjustments, and cross-protocol value capture—a long and challenging journey ahead.
DeFi Intelligence Roadmap: From Automation Tools to Agent Networks
In summary, we have witnessed the evolution of DeFi intelligence—from automation tools to intent assistants to intelligent agents.

The first stage is “Automation Infrastructure,” characterized by rule-based and conditional execution enabling basic on-chain automation—such as triggering trades or rebalancing based on time or price. Representative systems are typically low-level execution frameworks like Gelato and Mimic.
The second stage is “Intent-Centric Copilot,” emphasizing user intent expression and execution suggestions. Systems at this stage move beyond “what to do” to understand “what the user wants,” then suggest optimal execution paths. Projects like Bankr and HeyElsa exemplify this, reducing DeFi’s usability barrier through improved intent recognition and interaction.
The third stage is “AgentFi Agents,” marking the formation of closed-loop strategies and autonomous on-chain execution. Agents can autonomously perceive, decide, and act based on real-time market conditions, user preferences, and strategy logic—enabling true 7×24 non-custodial on-chain fund management. Simultaneously, AgentFi’s ability to manage user funds without step-by-step authorization raises significant debates around security and trust—becoming an unavoidable core issue in AgentFi design. Representative projects include Giza ARMA, Theoriq AlphaSwarm, Almanak, and Brahma, all demonstrating solid implementation capabilities in strategy deployment, security architecture, and product modularity—forming the backbone of today’s DeFi Agent landscape.
We look forward to the emergence of “Advanced AgentFi Agents”—not only achieving autonomous execution but also handling complex cross-protocol, cross-asset scenarios. This is our vision for the advanced form of DeFi intelligence:
-
Pendle Yield Trading: Future agents will fully manage maturity rollovers and strategy orchestration, maximizing capital efficiency.
-
Funding Rate Arbitrage: Cross-chain arbitrage agents will precisely capture every funding rate differential opportunity.
-
Swap Strategy Combos: Swap becomes a key node in multi-strategy yield pathways, enabling portfolio value leaps.
-
Staking and Restaking: Agents will continuously optimize staking portfolios, dynamically balancing return and risk.
-
RWA Asset Management: As real-world assets go on-chain, agents will allocate global stable-yield assets intelligently.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News














