As stablecoin yield strategies become increasingly diverse but user energy and execution efficiency are limited, AgentFi, as the next stage paradigm of the "AI + DeFi" integration, provides DeFi users with intelligent agency solutions that do not require monitoring and are automatically optimized.
Written by: 0xjacobzhao and ChatGPT 4o
The "Stablecoin Yield Guide" written last month detailed the characteristics, income sources and potential risks of eight stablecoin interest-earning models, including stablecoin lending (Lending & Borrowing), liquidity mining (Yield Farming), market-neutral arbitrage (Market-Neutral Arbitrage), US Treasury-Backed RWA Yields, option structured products (Structured Product), yield tokenization (Yield Tokenization), a basket of stablecoin income and stablecoin staking pledge income. It became a recommended reading for many Chinese media in April and had a certain communication influence in the Chinese Crypto circle. Many stablecoin teams contacted me and provided me with new information, views and perspectives. Most projects can still be summarized into the existing eight types of income models, and what is exciting is that I discovered a new paradigm of stablecoin interest-earning that is happening: AgentFi.
1. Classification and comparison of DeFAI and AgentFi
Although DeFAI and AgentFi are both in the intersection of "AI + DeFi", their technical essence, product form and paradigm mission are completely different. AgentFi is different from traditional DeFi or generalized DeFAI projects. Its core features are: AI Agent is the main executor and decision carrier, fully automatic decision-making and closed-loop execution, rather than simply using AI as a decision-making auxiliary tool for data analysis and strategy enhancement.
DeFAI category: Use AI to assist or enhance DeFi-related functions, such as tool-based applications such as strategy optimization, data analysis, and investment advice, and AI models assist human decision-making. For example, Kaito, Numerai, and Gauntlet, which focuses on DeFi yield optimization. This type of financial automation product relies on users to clearly specify instructions and strategies, and executes the process from "instructions to operations".
AgentFi category: Building a financial system with on-chain AI agents as the first executors. Agents can autonomously reason, make decisions and complete transactions or governance operations. Users authorize agents to execute on their behalf, and agents actively make decisions and execute. Using the "from intention to result" paradigm, users only need to express their goals, and the system can complete strategy formulation and execution.
At present, the market is flooded with a large number of Agent launch platforms or trading markets. Compared with AI Meme tokens that are purely hyped and have no practical use, there are only a handful of AgentFi projects that truly focus on the world of Crypto Native Defi. Through a lot of online research and screening, after eliminating projects that are too early or obviously empty, we have made the following simple classification and rating of the AgentFi projects currently on the market:
✅ Stablecoin revenue optimization
✅ Strategy combination/liquidity mining
✅ On-chain transaction execution
The above 9 AgentFi projects are all at the initial stage of having a closed-loop on-chain execution, clear scenarios, and basic products. But to be honest, the AgentFi track is still too early and the experience of most products is hard to describe. Compared with mature Defi protocols, the product gap is still large. We hope that the AgenFi track can get rid of the Meme token hype stage and enter the real PMF stage in the future.
2. Stablecoin Yield Optimization in AgentFi
The track of stablecoin yield optimization is not new in the DeFi world. From the earliest Yearn Finance (YFI) automated yield aggregator and Idle Finance's automatic asset rebalancing and strategy combination platform, to Gauntlet, a professional DeFi risk modeling and protocol-level yield optimization platform for institutions, they are all relatively representative automatic yield optimization projects before AgentFi.
AgentFi has initially realized the vision of mutual integration of Crypto and AI: AI enables users to better participate in the native Defi world of Crypto through on-chain information collection and decision-making capabilities, while taking advantage of the immutability of blockchain, transaction details are retained in the form of on-chain logs to ensure that logic is verifiable and behavior is traceable. As a continuation of the article on stablecoin income, we focus on the only target in the "stablecoin income optimization" category that has received five-star reviews in terms of product availability under the AgentFi track - ARMA.
3. Giza Project Vision and Paradigm Concept
Giza is a blockchain infrastructure project focused on building "Verifiable AI Agents" and is committed to promoting the emerging paradigm of Agent-native Finance. As one of the few project teams in the ZKML track, Giza is different from Modulus Labs or EZKL, two teams that are more inclined to theoretical research or infrastructure development. It pays more attention to engineering and practicality. It is the first project in the ZKML track to realize the complete closed loop of "on-chain intelligent agent + verifiable execution + AI scheduling".
Giza proposed a groundbreaking concept - Xenocognitive Finance, which is to introduce AI agents as a new interface in decentralized finance to solve the cognitive bottleneck of human participants and reconstruct market intelligence and structure. AI agents are not passive tools, but "cognitive representatives" with independent reasoning capabilities and on-chain execution permissions. They enhance the information processing capabilities of the entire market through a distributed network of AI agents, so that market intelligence expands with the scale of participation rather than collapses, and provides credible results through zero-knowledge proofs (ZK). In short, Xenocognitive Finance is a new paradigm that expands human cognitive capabilities through autonomous intelligent agents, achieves "cognitive offload" without sacrificing sovereignty in a non-custodial model, and efficiently participates in decentralized finance without having to constantly monitor the market or automatically switch protocols.
The architecture of Giza Protocol consists of the following three layers:
Semantic Abstraction Layer: Breaks through the semantic barriers between AI and blockchain, helping AI understand the blockchain world.
Decentralized Execution Layer: Provides a scalable, secure, and censorship-resistant agent execution environment.
Agent Authorization Layer: Ensures that agents operate in an unmanaged, secure, and controllable manner.
Giza's proxy execution process is completed by three core modules in collaboration and is divided into four key steps:
Request processing (semantic abstraction layer): The AI agent initiates an operation request (such as "deposit USDC into AAVE") through the MCP protocol. The system parses the natural language policy into standardized on-chain operation instructions, while verifying the format, semantics, and permission boundaries to ensure that the request is legal and executable.
Protocol interaction (decentralized execution layer): The execution node calls the DeFi protocol (such as lending, exchange) according to the instructions to complete the actual operation. The system will dynamically optimize the transaction path, taking into account gas costs, slippage tolerance and protocol liquidity. After the task is executed, the results and task proofs are generated for subsequent verification;
Execution verification (authorization layer + execution layer): The system checks whether the operation complies with the Session Key authorization scope (protocol type, amount, time) set by the user. Multiple Attester nodes independently verify the operation. Aggregator collects signatures to reach consensus to ensure that the results are secure and reliable.
Result feedback (semantic abstraction layer): The execution results are converted into structured semantic information (such as income changes, asset distribution) and returned to the AI agent system for subsequent strategy adjustments and automatic cycle decision-making.
In July 2023, Giza announced the completion of a $3 million Pre-Seed round of financing, led by CoinFund, with participation from StarkWare, TA Ventures and Arrington Capital (founded by the founder of TechCrunch). Angel investors include Rand Hindi and Julien Bouteloup. In May 2025, Giza completed another $2.2 million seed round of financing, led by Base Ecosystem Fund, with participation from CoinFund, Arrington Capital, Re7 Capital and Contango Digital Assets. These two rounds of financing demonstrate the capital market's continued optimism about Giza's vision in the field of "AI + DeFi".
4. Representative product ARMA: Stablecoin yield optimization agent
ARMA (Autonomous Revenue Management Agent) is a stablecoin revenue optimization intelligent agent launched by Giza, which aims to maximize the automated revenue across protocols for users. It has been deployed on the Base network and plans to expand to more Layer2 ecosystems, supporting mainstream lending protocols such as AAVE, Morpho, Compound, and Moonwell.
4.1 Overview of ARMA Working Mechanism
Intelligent Strategy Scheduling: Real-time analysis of the stablecoin returns (APR), transaction costs, and rebalancing opportunities of each protocol, automatically executing optimal rebalancing operations.
APR Optimization System: Through comprehensive optimization of interest rate comparison, cost assessment and position adjustment logic, it ensures that each operation brings net profit; the actual annualized return can be up to 2 times that of traditional static positions.
Automatic compounding mechanism: Automatically collect and reinvest incentive tokens, converting rewards into original stablecoins; the frequency of compounding is dynamically optimized based on position size and transaction costs to improve efficiency.
Smart currency exchange logic: Supports dual-currency operations of USDC and USDT. When necessary, it can automatically and efficiently exchange currencies through integrated DEX to ensure that users can always withdraw assets in the original currency, improving flexibility and experience.
4.2 Overview of ARMA component modules and security architecture
The Giza Agent architecture consists of six modules: Smart Accounts, Session Keys, Agent Core, Protocol Integration, Risk Module and Accounting Module. The modules work together to build a secure, efficient, intelligent and verifiable on-chain automated execution system.
At the same time, ARMA ensures the security of user funds and operations through a three-layer protection mechanism:
Smart Account Security: Based on self-hosted smart accounts, users always have control over their assets, and Giza cannot access private keys; the contract has been audited and supports flexible permission management.
Access control mechanism: Use Session Key to fine-tune proxy permissions, limiting them to specific protocols, operation types, and validity periods, to avoid the risk of full authorization.
Risk management system: Only carefully selected protocols are connected, focusing on stable profit scenarios; the entire transaction process is traceable to ensure transparent and auditable operations and enhance the overall security of the system.
4.3 ARMA charging mechanism and revenue management
ARMA’s fee and revenue mechanism is user-friendly, transparent and efficient, covering three aspects:
Fee structure: Billing is based on revenue, which is transparent and traceable. A 10% success fee is charged only on the actual revenue generated. This fee is settled once when the user withdraws money, and all fee calculation processes can be viewed in real time through the dashboard.
Reward management: Automatic collection and unified distribution. The ARMA agent will automatically collect and aggregate all rewards from various integrated DeFi protocols. All rewards will be returned together with the principal and income when the user withdraws, without the need for manual operation by the user.
Asset Management: ARMA does not charge any deposit or withdrawal fees. No matter how many protocols are involved or what tokens are involved, users always get their funds back in the original deposited tokens when withdrawing, ensuring the convenience of fund recovery.
4.4 ARMA’s access to Defi protocols and data performance
Through ARMA's official data dashboard (metrics.gizatech.xyz), we can intuitively observe the continued growth of its asset management scale and the number of smart agents within a few months of its launch. At the same time, the dashboard also shows the dynamic flow of assets between mainstream lending protocols such as Morpho, Aave, and Fluid. We believe that as ARMA accesses more DeFi protocols, the number of active agents increases, and the scale of funds continues to expand, its flywheel effect of positive growth will be further strengthened.
5. Token core positioning and incentive mechanism
At present, Giza has just announced the complete details of the token economic model, and TGE token issuance has become an important milestone worthy of attention. $GIZA is the native asset of the Giza smart proxy network, with immediate practicality and long-term value capture capabilities, and is based on community priority, product drive, and mechanism restraint as core design principles.
Summary of key points
Reasonable initial valuation: TGE's initial FDV is $70M, which is much lower than similar AI projects and has significant upside potential.
Zero unlocking startup: The team and investors will not unlock tokens during TGE, avoiding initial selling pressure and being highly consistent with the community.
The product has been launched: ARMA smart agents have managed over $30M in on-chain assets, and market demand is supported by real use cases.
The purpose of the token is clear: $GIZA is used for pledge verification, permission control, protocol governance and agent incentives, and is a core asset.
Incentives and security are bound together: Proxy nodes need to stake $GIZA, and malicious behavior will be punished to ensure safe and reliable execution.
Value return mechanism: Part of the agent operation income flows into AMM for $GIZA repurchase or redistribution, building a sustainable incentive closed loop.
Community-oriented distribution: Tokens are allocated to real users and participants first, emphasizing fair distribution and long-term co-construction of value.
Dual staking mechanism: Both node operators and ordinary users can stake $GIZA to obtain execution rewards and governance rights respectively.
Time-weighted design: The longer you stake, the higher the reward multiplier and governance rights, up to 5 times, to encourage long-term participation.
Progressive governance: The governance mechanism is advanced in stages, transitioning from the Security Committee to representative governance and professional committees, and ultimately achieving decentralized autonomy.
Giza Points Incentive System:
In order to encourage users and developers to participate in the construction of the Giza smart agent ecosystem, Giza launched the Giza Points System (GPS) at the end of 2014 as one of the bases for future native incentive distribution. The points system emphasizes "long-term contribution" over "short-term behavior" and is open to the following key participants:
User side: Continuously accumulate points through on-chain behaviors such as agent product usage, feedback, and stress testing;
Developers: Get rewards for technical contributions such as integrating new agents, optimizing execution logic, and submitting security improvement suggestions;
Community members: Participating in education promotion, governance discussions, ecological co-construction and other activities can also earn points.
Currently, ARMA is in the high token subsidy stage of the startup period. Users who deposit stablecoins can get a fixed annualized rate of return of 15% in the first 3 months, which is composed of basic interest and additional rewards. In addition to the fixed income, the first $10M deposit users can also receive a 1% share of the total $GIZA token supply in proportion, and the tokens will be released linearly within 9 months. No lock-up, no penalty, suitable for users who pursue both short-term stable income and the long-term potential of $GIZA.
Giza Growth Flywheel Effect (Flywheel)
Giza has built a growth flywheel driven by the practicality of intelligent agents, protocol usage, token value and ecological participation: intelligent agents improve the efficiency of user fund management and strategy execution, drive more DeFi operations and protocol interactions, generate continuous income back to stakers, enhance the value of $GIZA tokens, attract more nodes and developers to join, and further expand the capabilities of the agent network, ultimately achieving a positive cycle of "more use → higher value → stronger security → broader capabilities", and promoting Giza to become a scalable and trustworthy Agent-native financial infrastructure.
Giza token governance mechanism and community participation design
Giza's governance mechanism adopts a phased evolution model, combining professional committees and community representation to ensure that the protocol is gradually decentralized on the basis of security and stability. In the early stage, it will be led by the Security Committee, and the community will provide opinions through voting and discussion; in the medium term, it will open up on-chain governance of key matters; and finally, it will be governed by a professional committee elected entirely by the community. Token holders can participate in decision-making by entrusting representatives with technical or economic backgrounds, and enhance voting influence through a time-weighted staking mechanism, encourage long-term participation and shared responsibility, and create a safe, efficient, professional and democratic governance system.
VI. Investment Analysis Logic and Potential Risks
The author met the Giza team during the ETHcc in Paris in 2023. Giza initially entered the market with the concept of ZKML, and then gradually evolved into an AgentFi project with more engineering implementation and DeFi scenarios in 2024. After a lot of research and careful evaluation of the mainstream stablecoin income projects and the emerging AgentFi projects in the market, we can still believe that Giza's ARMA is one of the very few AgentFi projects with real product polishing and on-chain execution closed loops. This "product before token" path is in sharp contrast to the Meme coins that are full of concepts, lack practicality, and eventually evolve into air coins in the DeFAI craze. At the same time, we have to admit that compared with other channels for stablecoin asset interest, the AgentFi model is still extremely early, and the protocol has not yet been verified through a complete market cycle. It is still in the "early head mine" period and has high volatility and high experimental characteristics. For DeFi Degen users who are accustomed to exploring cutting-edge opportunities, they may use part of their stablecoin assets for trial, but they still need to pay attention to risk diversification.
Investment Logic
Paradigm innovation narrative: The Giza project is located at the intersection of AgentFi, ZK and AI technologies. It has created the original "Xenocognitive Finance" theory and proposed a future financial architecture that achieves "cognitive offloading" through on-chain intelligent entities. It has a clear first-mover advantage and philosophical depth.
Leading product practicality: ARMA launched by Giza is one of the very few AgentFi products in the current market that has been launched and has achieved a closed loop of automated agent execution. The product experience is better than most "air projects" that are still in the conceptual stage or purely token-driven.
Early first-mine profit window: Stablecoin income itself has relatively controllable risk attributes under the premise of protocol security. Combined with the token subsidy mechanism provided by ARMA at the current stage, it provides early participants with a "first-mine" opportunity.
Potential risks
It is still early days to build user trust: The AgentFi model is still in the exploratory stage, and the market's trust in on-chain AI agents is still limited. In particular, there are psychological barriers in terms of fund custody and transparency of strategy execution, and a user growth flywheel has not yet been formed in the short term.
Complex systems with superimposed security risks: The ARMA protocol integrates multiple highly complex modules such as AI reasoning, on-chain account abstraction, and multi-protocol integration. Its security depends on the dual correctness of contract design and reasoning logic. Once any link fails, it may cause asset risks.
The path of token value release is unclear: the current Crypto market is highly speculative, and token performance is more dominated by emotions and liquidity. The Giza team is not a speculative team, and the incentive structure, pledge mechanism and value return model of the $GIZA token still need further market verification, which puts higher requirements on long-term investors.
8. Conclusion
In the context of increasingly rich stablecoin yield strategies but limited user energy and execution efficiency, AgentFi, as the next stage paradigm of the integration of "AI + DeFi", provides DeFi users with intelligent agent solutions that do not require monitoring and are automatically optimized. This article, represented by Giza and its core product ARMA, systematically sorts out the development status, technical characteristics and profit potential of the AgentFi track. With "different cognitive finance" as the cornerstone of its concept, Giza has taken the lead in realizing a complete closed loop of AI strategy + non-custodial execution. It is one of the few AgentFi projects that currently has both conceptual depth and product practicality. ARMA focuses on the optimization of stablecoin returns. It has launched multiple chains and integrated mainstream lending protocols. With a structured token incentive mechanism, it has an early head mining effect.
Of course, we should also be aware that AgentFi is still in its early stages, and user trust, execution stability, and token value capture mechanisms are all waiting for further market verification. But for this reason, for users and builders who value product prospects and are willing to lay out early dividends of paradigm shift, AgentFi, especially Giza/ARMA, may represent the harbinger and opportunity of the era of intelligent financial institutions.
