CoinW Research Institute: Exploring the Present and Future of DeFAI

Key Points

• DeFAI addresses complex operations in DeFi and the lack of market risk prediction among investors. It is also seen as a more intelligent abstraction layer. DeFAI is expected to push DeFi towards a new stage of intelligence, opening up new opportunities for both institutions and individual users. Currently, DeFAI is still in its early experimental stage. This report analyzes 15 key representative projects in four main categories: operation simplification, market analysis, infrastructure, and portfolio management.
• In the operation simplification category, representative projects include Griffain, Hey Anon, Orbit, neur, and Slate. Griffain simplifies user operations and provides a comprehensive solution for DeFi, meeting users’ diverse investment needs. Hey Anon’s core functions include the Automate framework and the Gemma agent, with a $20 million AI agent fund support from DWF Labs. Orbit allows users to interact with multiple protocols across different chains, executing all operations through a simple AI agent request. Neur is an open-source full-stack application integrating LLM models. Slate is designed for advanced traders, using AI to analyze messages from Telegram, Discord, X, and on-chain sources for precise and fast trading. Slate has not yet issued a token and is currently in the trial registration phase.
• In the market analysis category, representative projects include Aixbt, VaderAI, and Acolyt. Aixbt is one of the most influential analytical AI agents with broad applications in smart investment research. VaderAI’s planned applications include autonomous hedge funds, influencer agencies, and on-chain arbitrage operations, with data analysis as its currently implemented use case. Acolyt is still in its early stages, focusing on automated social interactions, terminal services, and API access, with its primary function being precise market analysis.
• In the infrastructure category, representative projects include SwarmNode.ai, Mode, and Brahma. SwarmNode helps developers deploy AI agents on the cloud without requiring a server. Its goal is to become a one-stop shop for AI agents, with each agent working independently and collaborating with others. SwarmNode has joined NVIDIA’s Inception Program, gaining access to more technical resources. Mode has transitioned from a modular Layer2 focused on DeFi to a DeFAI project, innovating at the interface, data, and infrastructure levels. Brahma facilitates AI agents in transitioning from analysis to execution, including trading and cross-chain DeFi operations. Brahma has received investment from top VCs, including Framework Ventures, Lightspeed Venture Partners, and The Lao, but has not yet issued a token.
• In the portfolio management category, representative projects include Flork, The Hive, Whispers, and Cleopetra. Flork provides DeFi functions such as sentiment tracking, social content analysis, token tracking, and LP analysis. It integrates staking, NFTs, and memecoins, making it a notable project in the Solana ecosystem. The Hive, the first-place winner of Solana’s hackathon, focuses on automating DeFi operations, offering market sentiment analysis, optimization strategies, and risk management models. Whispers is an open-source protocol that allows developers to build, orchestrate, and embed AI agents. It launched in January 2025 and has rapidly evolved, receiving support from the Solana Foundation. Cleopetra, another Solana hackathon winner, manages LPing on Solana DEXs through AI agents. The project is currently in its Beta phase, accessible only to a limited number of testers, and has garnered strong community support, including attention from Solana co-founder Toly.
• Most current DeFAI projects are still in their early exploration stages and require further development to address real-world problems and enhance user experience. However, DeFAI’s proposal to integrate AI with DeFi and improve DeFi’s narrative may mark a new beginning. This report predicts that the future of DeFAI will involve AI-driven intelligent economic models, agents that directly convert intentions into actions, modular collaboration trends, and specialized applications emerging as valuable opportunities.

CONTENT

• Key Point
• I. A New Direction for AI: DeFAI
• 1.What Problems Does DeFAI Solve?
• 2.A More Intelligent Abstraction Layer?
• II. Key DeFAI Projects
• 1. Simplified Operation Category
• 1.1 Griffain
• 1.2 Hey Anon
• 1.3 Orbit
• 1.4 neur
• 1.5 Slate（No Token）
• 2. Market Analysis Category
• 2.1 Aixbt
• 2.2 VaderAI
• 2.3 Acolyt
• 3. Infrastructure
• 3.1 SwarmNode.ai
• 3.2 Mode
• 3.3 Brahma（No Token）
• 4. Portfolio Management
• 4.1 Flork
• 4.2 The Hive
• 4.3 Whispers
• 4.4 Cleopetra（No Token）
• III. DeFAI Outlook

1. AI-Driven Intelligent Economic Models

2. AI Agents Transition from Intent Layer to Execution Layer

3. Modular Collaboration as a Development Trend

4. Niche Innovation Applications as Value Hotspots

I. A New Direction for AI: DeFAI

1. What Problems Does DeFAI Solve?

Recently, market sentiment has been negative. After experiencing short-term hype, the AI sector has gradually returned to rationality, shifting its focus from speculation to practical applications. Investors are now paying attention to AI projects that solve real-world problems, with DeFAI emerging as a key trend. DeFAI is a fusion of DeFi and AI, primarily enhancing DeFi’s intelligence and efficiency through AI. It provides new solutions for DeFi, addressing complex operations and insufficient market risk prediction by investors. DeFAI is expected to propel DeFi into a new phase of intelligence, offering new opportunities for institutions and individual users. Currently, DeFAI’s core applications include smart investment, automated trading, trustless lending, and security risk management.

2. A More Intelligent Abstraction Layer?

DeFAI emphasizes intent-driven operations, simplifying DeFi interactions, similar to chain abstraction. However, DeFAI enhances the existing chain abstraction concept. Chain abstraction primarily focuses on eliminating users’ awareness of underlying blockchain layers, making cross-chain interactions and interoperability protocols more seamless and efficient. In contrast, DeFAI integrates AI agents with intent-driven mechanisms, allowing users to execute optimal trading paths without manually selecting the best on-chain route. With DeFAI-powered products, users can achieve optimal cross-chain asset management and trading execution through smart contracts or natural language instructions.

As a result, DeFAI is seen as a more intelligent abstraction layer. According to CoinGecko, there are currently 82 DeFAI projects, mainly concentrated on the Solana and Base chains. Most DeFAI projects are still in the early exploration stage. The following sections will analyze 15 representative projects at this stage.

II. Key DeFAI Projects

1. Simplified Operation Category

1.1 Griffain

Griffain simplifies user operations and provides a comprehensive solution for DeFi, including token/NFT creation, arbitrage trading, whale tracking, and liquidity mining, catering to users’ diversified investment needs. In terms of trading, it supports token swaps using natural language. In terms of services, it allows for customizable user services and helps users accurately track whale movements. Griffain’s DEV @tonyplasencia3 is one of Solana’s core developers. Solana co-founder Toly views Griffain as the first on-chain ChatGPT for Solana. Additionally, Griffain has received official recognition from Solana.

Griffain’s application is already available on mobile, but its functionalities are similar to the web version, mainly offering basic services. Recently, Griffain launched Agent Studio, integrating data from Cookie DAO, allowing users to receive the latest market data via email. Currently, Griffain users participate through invitation, with two main access methods: Early Access Pass and Griffain Pass. The Early Access Pass was released on November 23, 2024, priced at 1 SOL, and sold out within a week. The Griffain Pass, currently the main participation channel, is priced at 2 SOL with no quota limit. Users purchasing the Griffain Pass receive a 0.5 SOL reward in their proxy wallet.

Most transactions within Griffain are conducted in SOL, and the functionality of the $GRIFFAIN token remains unclear. Its primary revenue sources include transaction fees and proxy service fees. Transaction fees consist of a base fee (0.0015 SOL per transaction) plus a variable fee. Proxy service fees include 0.25 SOL per 250 tweets, 0.1 SOL per 100 emails, and 0.1 SOL per 100 generated images. Griffain enables users to execute most on-chain operations via natural language, lowering the barrier to entry. In the future, users may only need a mobile app to execute complex on-chain investment strategies. Recently, Griffain has integrated with Backpack and pump.fun, allowing AI agents to facilitate token transactions.

Griffain’s native token is $GRIFFAIN, with a current market capitalization of $64.92 million.

1.2 Hey Anon

Hey Anon is an AI-driven DeFi protocol that simplifies operations in two key ways: first, by automating complex DeFi interactions, including cross-chain transactions, swaps, staking, and lending; second, by aggregating and analyzing social data from Twitter, Telegram, Discord, GitHub, and Gitbook to provide users with better investment strategies.

Hey Anon’s core functionalities include the Automate Framework and the Gemma AI Agent:

• Automate Framework: Supports parallel integration, allowing hundreds of pull requests to simultaneously add new “deterministic paths,” simplifying user operations. Built on TypeScript, its core goal is to serve as a multi-chain abstraction layer. Currently, it supports 18 blockchains, including ETH, BSC, and Base, integrating over 50 protocols across lending, staking, DEXs, bridges, wallets, aggregators, data infrastructure, AI, and stablecoins.
• Gemma AI Agent: Integrates social data from Twitter, Telegram, Discord, GitHub, and Gitbook to provide real-time market insights and analysis, helping users make better investment decisions.

Source:@HeyAnonai

Hey Anon’s founder, Daniele Sesta, also founded the Spell project, which once reached a market capitalization of $2 billion. Hey Anon has partnered with DWF Labs and received $20 million in AI agent funding support from DWF Labs. The $ANON token is primarily used for Anon DAO governance, AI agent services, and participation in project development decisions.

Currently, Hey Anon’s product remains in its early stages. Users can connect to a bot, such as on Telegram, and issue commands in natural language to complete operations. The latest version, HeyAnon V0.2, has introduced more reasoning models and integrated @magpieprotocol and @VenusProtocol. Additionally, Hey Anon’s iOS and Android apps are launching soon.

Hey Anon’s native token is $ANON, with a current market capitalization of $98.83 million.

1.3 Orbit

Orbit is an AI agent launched by @sphereone_ (a Coinbase-invested crypto payment platform). It primarily addresses DeFi investment needs, including cross-chain transactions, staking, and lending. Orbit allows users to interact with multiple protocols across different chains using a simple prompt, with the AI agent executing all operations automatically. The project has been in development for over a year and has integrated over 100 blockchains and thousands of DeFi protocols.

Orbit’s token is mainly used for accessing premium platform features, community events, and product upgrades. Additionally, 30–50% of Orbit’s net revenue is allocated to public market buybacks and rewards.

Users only need to register an account to start using Orbit. Currently, four agents are available, but they remain at a basic conversational stage. Orbit’s AI agent is still in its early phase, unable to complete complex operations like fund transfers or provide accurate whale position analysis. For example, when CoinW Research attempted to analyze content on X using Orbit’s agent, the investment suggestions provided lacked real value.

Orbit’s native token is $GRIFT, with a current market capitalization of $17.03 million.

Source：chat.orbitcryptoa

1.4 neur

neur is designed for the @solana ecosystem, providing an intelligent interface for seamless interaction with DeFi protocols, NFTs, and more. Its main feature is an open-source full-stack application that integrates an LLM model. The neur ecosystem has already integrated multiple protocols: in the DeFi sector, it has integrated Jupiter; for token issuance platforms, it has integrated Pump Fun; for the NFT market, it has integrated Magic Eden; and for market analysis, it has integrated DexScreener.

90% of neur’s native token, $NEUR, is already in circulation, with 3% allocated to the project fund and 7% permanently burned. The token is primarily used to incentivize users to participate in platform development. Currently, neur is still in its Beta testing phase. Early users need to pay 1 SOL to participate in testing (later, it will support its native token $NEUR). At this stage, the product is still in its infancy — on-chain transfers and transactions are not yet functional, and on-chain analysis remains underdeveloped. In the recent EAP testing event, users needed to hold 20,000 $NEUR to participate. However, the community has expressed skepticism about the product’s quality, with many holders believing that neur has not delivered much so far.

The native token of neur is $NEUR, with a current market capitalization of $4.71 million.

Source：neur.sh

1.5 Slate (No Token)

Slate is a general-purpose AI agent that executes on-chain operations with a single command by converting natural language instructions into blockchain actions. Currently, Slate primarily addresses the trading needs of advanced traders by analyzing messages from Telegram, Discord, X, and on-chain data to enable precise and rapid trading.

Slate has integrated Solana, Base, and Hyperliquid chains and provides MEV protection and optimized trading paths. Designed for speed and precision-oriented traders, Slate helps users streamline their trading processes and eliminates tedious manual operations.

Slate has not yet issued a token, and its token utility remains unclear. The main products of Slate are in the application registration trial phase, and users can only register through an invitation from the project team. Currently available agents include wallet tracking and Telegram token monitoring, though their use cases are limited. Despite being in the early development stage, Slate’s interface is relatively user-friendly, with clear user guidance and a well-structured roadmap, making it worth further observation.

Source：docs.slate.ceo

2. Market Analysis Category

2.1 Aixbt

Aixbt is an AI agent built on the Base chain, developed by the Virtuals Protocol team, focusing on on-chain data analysis. Its main functions include:

• Extracting market information and monitoring real-time DeFi ecosystem data, such as trading liquidity, yield fluctuations, and fund flows.
• Providing KOL-related services by analyzing market insights from key opinion leaders (KOLs) to help users assess market trends.
• Offering intelligent DeFi investment advice based on on-chain data and AI calculations, optimizing investment strategies for better returns.

Currently, Aixbt is one of the most influential analytical AI agents, widely applied in intelligent investment research. Other functionalities are still under observation. Users can mention @aixbt_agent on X to interact with the bot for analysis. Holding 600,000 $AIXBT tokens grants terminal privileges, allowing access to more precise market analyses. The token’s main functions include governance rights and unlocking platform privileges.

The native token of Aixbt is $AIXBT, with a current market capitalization of $180 million, ranking first in market cap within the DeFAI sector.

2.2 VaderAI

VaderAI is an AI agent on the Base chain (also expanded to Solana) with the primary goal of developing a comprehensive ecosystem where both AI agents and humans can establish and manage their own DAOs, fostering a diverse agent-driven ecosystem.

VaderAI features three core products:

• Vader Fun: A DAO aggregation platform that allows users to participate in DAOs managed by both AI agents and humans.
• VaderAI: The DAO managed by VaderAI’s autonomous trading agents.
• VaderAI KOL: Focuses on managing DAOs for KOLs on the X platform.

The most crucial component is Vader Fun, which consists of two types of DAOs:

• Passive DAOs, managed by VaderAI, charge a 0.5% management fee, with revenues used to reward $VADER stakers.
• Active DAOs, run by individual managers, charge a 0–20% performance fee, with earnings primarily used to reward $VADER stakers.

DAOs of VaderAI Source：vaderai.gitbook.io

Future applications of VaderAI include autonomous hedge funds, KOL agent networks, on-chain arbitrage, NPC guilds in video games, customer support solutions, content creation, and entertainment. Currently, the main implemented use case is on-chain data analysis, with other functionalities still under observation.

The main functions of the VaderAI token include launching DAO-related agents and staking rewards. Users must stake a certain amount of $VADER to initiate a DAO. The staking section has a total of 530 million $VADER staked, with a minimum staking period of 30 days.

The native token of VaderAI is $VADER, with a current market capitalization of $15.37 million.

2.3 Acolyt

Acolyt launched in December 2024 and remains in its early stages. The project focuses on automated social interaction, terminal services, and API access, each with different requirements:

• Automated social interaction: Free for users.
• Terminal services: Requires token staking.
• API access: Requires subscription or unlocking via reward points.

Acolyt’s team has extensive hands-on experience, with core member @kurorosage also being a key contributor to @Vader_AI_. The project has a well-developed ecosystem plan, covering aspects such as data collection and framework implementation.

Source：@condzxyz

Acolyt’s main feature is its precise market analysis capabilities, characterized by:

A lightweight open-source framework called Signal Framework, enhancing AI agents’ human-like perception of user needs.

A goal-driven autonomous evolution approach.

Optimal path selection for service delivery, improving resource efficiency. For example, it uses DeepSeek-R1 for detailed posts and GPT-Mini for quick posts.

Acolyt accepts $USDC and $USDT for payments, with discounts available for using the native token $ACOLYT. The token is mainly used for staking rewards, unlocking terminal and API services. Revenue distribution allocates 80% to operations and 20% to token buybacks. The token distribution includes 75.55% for the community, 12.2% for the team, and 12.2% for the treasury.

Notably, Acolyt’s whitepaper states that its business model revolves around seamless integration with clients through its terminal and API. Its economic growth flywheel also includes B2C and B2B-driven expansion. However, the current delivered product mainly focuses on X-based social media interaction, with the roadmap extending to the third phase, and other areas yet to be implemented.

The native token of Acolyt is $ACOLYT, with a current market capitalization of $8.72 million.

3. Infrastructure

3.1 SwarmNode.ai

SwarmNode is an upgraded version of BotFleet, launched in December 2024, focusing on creating “swarms” of autonomous AI agents. SwarmNode enables developers to deploy AI agents in the cloud without requiring a server. It aims to be a one-stop shop for AI agents, where each agent can operate independently and collaborate with others. Users can deploy their AI agents easily through SwarmNode’s Python SDK or REST API.

Source：@swarmnode

SwarmNode has introduced service fees for its agents, offering 12 hours of free service upon registration. Afterward, users are charged based on usage, with high-quality agent services costing only $0.50 for 10 hours. The SwarmNode token is primarily used for paying service fees, community governance, and ecosystem development. Its main user base includes entrepreneurs, small and medium-sized enterprises (SMEs), and AI enthusiasts. Currently, SwarmNode offers only a limited number of agent services, with just eight available in the standard bounty service, while more are in development.

Additionally, SwarmNode has joined the NVIDIA Inception program, which provides startups with cutting-edge technology tools, resources, and opportunities to connect with investors. Its collaboration with leading Web2 AI enterprises increases the likelihood of successful product implementation.

Source：@swarmnode

SwarmNode’s native token is $SNAI, with a current market cap of $29.21 million.

3.2 Mode

Mode transitioned from a modular Layer2 focused on DeFi to a DeFAI project and was launched in May 2024. Mode’s vision is to expand the boundaries of DeFi by enabling AI-to-AI interactions, where smart agents autonomously execute complex financial tasks.

Mode has integrated 7,848 AI agents and innovates within DeFAI on three levels:

• Interface Layer: The Mode AI terminal allows users to manage an integrated on-chain portfolio. This feature is currently available primarily to veMODE holders.
• Data Layer: Mode utilizes the Synth Bittensor subnet, which leverages Bittensor to create powerful synthetic predictive data for AI agents and LLMs. The testnet API is currently live, and Synth aims to expand from single-asset price paths to multi-asset price paths and eventually Monte Carlo simulation databases for various industries.
• Infrastructure Layer: Mode’s L2 infrastructure is continuously upgrading to ensure secure and cost-effective interactions between agents and humans. Initial upgrades include an AI security sequencer and Superchain Interop, both currently in the testnet phase.

Source：@JRossTreacher

Token Distribution:

• 35% allocated to user and developer airdrops
• 19% for investors
• 19% for core contributors
• 27% for the foundation and on-chain treasury

The $MODE token is used for governance, DeFi integration grants, community incentives, user and developer activities, and ecosystem development.

Mode is a key infrastructure project in the DeFAI sector. Its native token, $MODE, has a current market cap of $15.91 million.

3.3 Brahma (No Token)

Brahma focuses on transitioning AI agents from analysis to execution, including trading, cross-chain operations, and other DeFi tasks. It enables secure smart accounts, customizable strategies, and multi-layer security checks. Brahma is built on ConsoleKit, allowing for more intelligent AI agent operations.

Brahma offers two types of accounts:

• Brahma: Designed for individual users seeking simplicity.
• Brahma Pro: Tailored for advanced users, on-chain teams, and asset managers. It supports multi-signature operations, access control, and sub-account management to mitigate risk.

Source：docs.brahma.fi

Brahma is still in early testing, with ConsoleKit’s alpha version set to launch soon. The platform integrates dApps, wallets, and other tools to enhance user experience.

Brahma currently supports multiple blockchains, including Berachain, Ethereum, Arbitrum, Base, Swell, and Mode. Its latest product update includes the Surge & Purge Agent, built on ConsoleKit, which dynamically adjusts trades to maximize profits. However, product testing indicates that usability remains a challenge, with slow asset transfers and limited operational efficiency, suggesting it is still in its early stages.

Brahma has not yet issued a token but has received investment from top-tier VCs, including Framework Ventures, Lightspeed Venture Partners, and The Lao.

4. Portfolio Management

4.1 Flork

Flork is an AI agent for managing liquidity pools, inspired by webcomics and positioned more as a Memecoin. The project has a CTO and integrates staking, NFTs, and Memecoin mechanics, making it one of the more popular projects in the Solana ecosystem.

Flork’s DeFi functionalities include:

• Sentiment tracker
• Social content analysis
• Token tracking
• LP analysis

However, its implemented features remain in early-stage development. The project has a strong community following, with active marketing efforts and quick responses to user feedback.

Staking: Currently supports FLORK and SOL, but both pools are at full capacity.

NFT Minting: A total of 8,888 NFTs are available, with 3,032 minted so far. Users must send 1 SOL to a designated address to mint an NFT.

Community Sentiment: Users show greater interest in the FLORK token itself rather than its derivative features.

Flork’s native token is $FLORK, with a current market cap of $17.78 million.

Source：florkcto.io

4.2 The Hive

The Hive, the first-place winner of the Solana hackathon, addresses automation challenges in DeFi. It features:

• Market sentiment analysis
• Optimized strategy execution
• Risk management models

Its two key advantages are:

Smart contracts streamline DeFi operations, reducing complexity.

Risk models allow for flexible DeFi strategy combinations, catering to different investor needs.

The Hive features two types of agents:

Specialized agents: Handle tasks like trading, staking, and analytics.

Advanced agents: Coordinate specialized agents to complete complex workflows.

Users simply input queries, and the network intelligently selects the best agent for the task. This modular architecture enables seamless scalability and adaptability.The Hive is currently free to use, requiring only an account login. The platform continues to improve, with one-click deposits, LP pool selection, and a natural-language token creation tool in development.The Hive has integrated Deepseek R and offers features like X-based market sentiment analysis, token tracking, and liquidity monitoring. It has 25,000 weekly active users, according to official data.

The Hive’s native token is $BUZZ, with a peak market cap of $200 million and a current valuation of $15.41 million.

Source：@askthehive_ai

4.3 Whispers

Whispers is an open-source protocol that allows developers to build, orchestrate, and embed AI agents.

Token Distribution:

• 65% to the community
• 20% to a locked liquidity pool
• 15% to the foundation

Whispers is already integrated with Gemini 2.0 Flash, deepseek_ai, and OpenAI. Early use cases include sending crypto to phone numbers and creating wallets via X accounts or phone numbers, potentially enabling DeFi transactions without traditional wallets.

Launched in January 2025, Whispers has seen rapid development, reaching 10,000 users within 24 hours and upgrading to Whisp v.2 within a week. It has gained support from the Solana Foundation and attracted attention from Solana co-founder Toly.

Source：@aeyakovenko

Whispers’ native token is $WHISP, with a market cap of $11 million.

4.4 Cleopetra (No Token)

Cleopetra, another Solana hackathon winner, launched in December 2024. It simplifies DeFi for everyday users by managing Solana DEX LPing via AI agents.

Users can add liquidity via Telegram, selecting a token and LP strategy, while the agent optimizes positions and minimizes impermanent loss.

Source：@cleopetrafun

Cleopetra is currently in Beta testing and is integrated with Solayer Labs and MeteoraAG. It is one of the few DeFAI projects making real technical advancements, particularly in LPing automation.

Cleopetra has strong community support and has also caught the attention of Solana co-founder Toly.

III. DeFAI Outlook

DeFAI’s core objective is to leverage AI-driven automated decision-making to enhance market prediction accuracy, optimize risk management, and provide users with more efficient and cost-effective financial services. DeFAI has a wide range of practical applications, including simplifying operational systems, improving DeFi efficiency, and enhancing security. However, most DeFAI projects are still in the early exploratory stage and require further development to solve real-world problems while optimizing user experience.

DeFAI is a novel concept, and we should distinguish its current state from its future potential. The combination of DeFi and AI to improve user experience and increase profitability could mark a new starting point for the market. This report believes that, in the future, DeFAI will evolve toward AI-driven intelligent economic models, where more agents can directly translate intent into action. Modular collaboration may emerge as a new trend, enabling smarter and more efficient DeFAI solutions. Furthermore, the segmentation and specialization of the sector will become key development trends, allowing niche applications to better meet market demands and become value hotspots.

1. AI-Driven Intelligent Economic Models

A well-structured economic model is a crucial foundation for a project’s sustainable development. This report suggests that sustainable DeFAI projects should feature more intelligent value capture, inflation and deflation mechanisms, and liquidity incentive mechanisms within their tokenomics.

• Value Capture: Users pay a certain fee when using AI agents, while those who contribute AI-optimized data can receive corresponding incentives, forming a closed-loop ecosystem.
• Inflation and Deflation Management: AI agents allocate different strategies at different stages of project development. In the early stage, an inflationary model encourages more users to participate in AI model training. In the mid-to-late stages, token buybacks and burns control the total token supply, continuously adding value to the tokens.
• Liquidity Incentive Mechanism: An intelligent liquidity mining mechanism can be employed, where AI monitors protocol capital flows and dynamically allocates token rewards to guide liquidity to where it is most needed.

In short, DeFAI’s tokenomics should be AI-driven, intelligently guiding user participation and promoting the project’s sustainable growth.

2. AI Agents Transition from Intent Layer to Execution Layer

Currently, most AI agents operate at the intent layer, where users express their desired actions through chat interfaces. However, as the sector develops, more agents will be able to directly translate intent into action. Projects like Brahma have already begun initial explorations in this direction, and more practical agents may emerge in the future.

This shift means that AI agents will no longer function solely as intermediaries for information transmission but will actively execute tasks, improving user experience and enhancing DeFAI efficiency.

3. Modular Collaboration as a Development Trend

In the DeFAI landscape, collaboration between AI agents is a key trend in improving DeFi transaction efficiency and user experience. Currently, most AI agents operate independently, attempting to provide end-to-end services across information analysis, investment decision-making, and trade execution. However, as market demand grows and technology matures, future DeFi AI agents may adopt a modular collaboration approach.

This means that multiple AI agents, each specializing in different tasks, will work together to build a more efficient and intelligent DeFi trading system.

4. Niche Innovation Applications as Value Hotspots

As an emerging field, DeFAI is still in its early development stages. With continuous technological advancements, the segmentation and specialization of this sector will become key trends in future development.

Currently, most DeFAI applications remain at the level of general-purpose tools and interfaces. For example:

• Analytical applications largely rely on generic analytical frameworks.
• Trading applications primarily depend on text-based interactions.

These generalized applications fail to fully meet the personalized needs of specific industries and user groups. Therefore, as market demand continues to evolve, more niche applications tailored to specific scenarios and needs will emerge. Customized DeFAI solutions designed for particular use cases will gain a larger market share.

These early-stage, niche-sector innovations may offer significant value opportunities.