AI Agents 2024 Review and Outlook: Where Have We Come From and Where Are We Going?

Original article from 0x Jeff

Compiled by Odaily Planet Daily Golem ( @web3_golem )

Review of AI Agent Development in 2024

2024 is a revolutionary year for AI Agents. About three months ago, terminal de vérité attracted everyones attention with its humorous personality, conversational style, and interaction with A16z co-founder Marc Andreessen , becoming the first millionaire agent and setting off a trend for AI Agents.

Soon after, Virtuals entered the space, pioneering “Agent Jetonization” and solidifying this narrative. Since then, innovation has exploded:

• Luna : This proxy launched an on-chain wallet reward fan function, which can now browse Twitter, analyze posts, and even join Google Meet.

• Conversational agents on Twitter/X: Some agents become trolls, while others focus on obtaining and sharing alpha information. For example :

1. aixbt : Known for its concise, actionable insights and some light spoofs;

2. Dolos : A sharp personality who has now developed his own framework to provide support for other agents through Dolion .

At the same time, AI Agents are gradually becoming more entertaining, with 3D models, voice functions, and cross-platform capabilities . Representative agents are as follows:

• AVA et Holoworld AI : the first 3D audio and video framework that gives agents 3D bodies, voices, and deeper personalities;

• zerebro : music agency that has released top albums and is about to launch its own framework ZerePy to enable more people to build agencies like Zerebro;

• Nebula : A Meme AI KOL that can create meme images and videos that appear in AR/VR environments and games;

• LucyAI : The first realistic anime agent that can speak multiple languages, live stream and interact with fans;

• DO KWEEN : A film agency that produces Netflix-quality dramas every week.

AI Agent Narratives in 2024

At the same time, ai16z and the open source innovation movement also gained traction, especially after the launch of the Eliza framework. Developers joined together to create toolkits, plugins, and other features to promote collaboration and innovation. During this period, Virtuals grew into a unicorn company, further consolidating its leadership in AI Agent distribution platforms.

The open source innovation movement has sparked the interest of the developer community and kicked off the biggest community collaboration of the year. Currently, more and more projects emphasize the importance of open source frameworks. As agents continue to evolve, new narratives have emerged to promote more agent collaboration:

• Agent Metaverse : Pioneered by Realis , which creates a Minecraft-style replica of Earth to house these AI agents, allowing them to interact and build a civilization.

• Gamification of agents : ARC Agents is a representative of this field, with artificial intelligence x games with reinforcement learning. Combining AI with reinforcement learning of games, it launched a game similar to Flappy Bird, which pits agents against each other, and the community contributes game data to help these agents grow. ARC recently revealed its vision for moving towards AGI.

• Cluster/collective intelligence : FXN is a representative in this field, aiming to establish a unified economy for AI agents. The idea of cluster is to let AI agents work together to achieve common goals. Virtuals is also promoting inter-agent interaction (or commercialization), which is a communication protocol that allows agents to seamlessly provide services to each other. At the same time, Story also announced the launch of an inter-agent communication protocol for IP, allowing agents to tokenize, monetize, and buy/sell/trade IP.

In parallel with these narratives, we can also see:

• On-chain trading agents: Originally proposed by Spectral , their Syntax v2 allows users to launch trading agents capable of trading on Hyperliquid. They have been dominating the field, but progress has been temporarily stalled due to a small bug. Another agent worth watching is Big Tony , which uses Allora s machine learning price prediction model to automatically trade mainstream currencies.

• InvestmentDAO : Initially ai16z was the representative, but now more DAOs are emerging, such as AIrthur Hayes et Aimonica . The general narrative is that these DAOs raise SOL on daos.fun (or other platforms) and use these funds to trade and invest for profit. If you can use the name of a Crypto VC or a well-known figure, then the InvestmentDAO you create will be more attractive.

• Defi Agent : Led by Mode , it is the preferred ecosystem for DeFi agents. The main application scenarios include AI-driven stablecoin mining, liquidity provision, lending, etc. High-quality teams in the ecosystem including Giza , Olas , Brian , Sturdy , QuillAI Network , etc. participated in the construction.

• AI App Store : ALCHEMIST AI is a representative in this field, which provides a no-code tool that allows users to create applications. MonShell is another AI App platform with a larger developer and user base, especially in the Web2 field.

• Abstract layer : Griffain et Orbit are representatives in this field, providing a chain-like abstract experience for all content on the chain, making it convenient for users to operate on the chain, especially very friendly to ordinary users.

• On-chain VC agent : Sekoia Virtuals wants to become a first-level rubber stamp for high-quality agent projects. Currently, it only strictly screens and invests in three projects, setting a precedent for on-chain VC.

• Other narratives : Freysa ’s on-chain puzzles, JailbreakMe ’s proxy cracking bounty rewards, H4CK Terminal ’s white hat AI, god et s8n ’s unique proxy models, representing God and Satan in a dialogue. More interesting are the agents focused on Alpha analysis, such as Rei (quantitative analyst), kwantxbt (TA analyst), and Nikita (general alpha analyst). Then there is Fartcoin, a suddenly popular meme project that even appeared on Stephen Colbert’s show and broke through the $1 billion market value. AI meme is being accepted by the public.

Data and framework development

• Cookie DAO is becoming the primary source of AI agent data and social metrics, and people in the industry rely on it to track agent influence, market value, and performance;

• Masa Integration with Virtuals to provide agents with real-time data, enabling them to self-learn and self-improve;

• TAOCAT is the first virtual proxy powered by the Bittensor subnet, demonstrating the possibility of real-time data (it is the only proxy token that has skyrocketed while other proxy tokens have fallen);

• AgentTank presents a framework that brings agents to computers, making them fully computer-operable so that they can interact entertainingly and provide interesting commentary on the Internet.

Other new frameworks:

• arc : A Rust-based RIG framework that has attracted attention for its versatility;

• Dolion : Evolved from Dolos , it is a toolkit for creating unique agents.

What have you learned from 2024?

The above may have overlooked some small narratives or AI agents, but through the development of AI agents, we can learn the following from this year:

Top teams with valuations over $50 million have their own fine-tuned models

They start by demonstrating the uniqueness of their agent’s use cases and models, and then provide a no-code framework for others to have an agent as good as their flagship agent, which in turn leads to higher agent value and pushes up agent token prices.

But this doesnt mean that you are advised to build your own framework or not to build on top of any other framework like Virtuals GAME and ai16z Eliza. If you dont have enough AI resources or capabilities, you should join these communities because with the help of tools you can quickly implement your ideas and experiment. At the same time, you should use Virtual and ai16z for distribution/marketing because these two places currently provide the best visibility and it is définitely positive EV to integrate and cooperate with them.

Investing in an agent with its own framework or the entire AI agent ecosystem will bring a better risk-reward ratio

If they manage to create a framework that people are willing to pay to build proxies for, that means there is enough interest and demand for the framework to drive or maintain the price. Arc is a great example of this, the first Rust framework quickly became popular and the price went up.

On-chain and DeFi applications will be the product-market fit (PMF) for crypto IA

The areas I think are currently delivering the most value include:

• Abstraction layers help people navigate the chain;

• Alpha agents share high-quality alphas, and people can make money from these alphas;

• Execution agents can help streamline trading, mining, providing liquidity, and loan execution.

Perhaps well see an agent that combines alpha discovery + execution soon.

Data is an integral part of every agency

Bad data = bad output. If data is gold, then data platforms like Cookie DAO are actually gold mines. Vana is an interesting L1 that tokenizes data into data liquidity pools. They have a DataDAO model that helps people co-own data, bring in data, and clean up data for AI Agents. Although the token economics may have problems, the product is very interesting.

Outlook for AI Agent Development in 2025

So far, we’ve explored where AI agents stand in 2024, reflecting on milestones and innovations from this year. Now, we’ll look ahead to 2025 — a year I believe AI agents will not only become more useful, but will begin to reshape the way we think about autonomy, intelligence, and collaboration.

Laying the foundation for 2025

Before we move on to the next step, it is worth emphasizing that Virtuals will continue to consolidate its position as the first distribution network for AI Agents on Base. Virtuals has become the preferred platform for agents to pair their liquidity, increase visibility, and form deeper cooperation with other high-quality projects. The total market value of Virtuals agents is currently about 3 billion US dollars, accounting for 77% of the entire AI Agent field (data source: Cookie DAO )

As more and more unique agents are created on Virtuals, and as these use cases become more diverse, more developers will be attracted to the Virtuals platform, regardless of whether they already have a token. This growth will also drive the price of the VIRTUAL token up.

While ai16z Dao has led the open source innovation movement with its Eliza framework, it currently lacks a launchpad and its token economics do not accrue the same level of value as Virtuals. Despite this, the future is still full of potential. They have recently formed a working group to improve their token economics, and a future launchpad could push ai16z to become the number one distribution platform on Solana, surpassing existing launchpads (if they decide to launch one).

We will also see top agents with product-market fit (PMF) gain significant capability upgrades in 2025. For example, AIXBT, which already leads the field in alpha-focused conversational agents, will likely further solidify its position with sharper responses and more insightful analytics.

This evolution will be reflected across the ecosystem as leaders in other verticals emerge, leading the way with their unique expertise and innovation.

What will be the trends in 2025?

2025 will be the year of professionalization of AI Agents. We will see the emergence of leaders in various verticals, each of which will dominate its own market segment:

• 3D Models : Proxies with high-quality visual designs suitable for games, AR/VR, etc.

• Speech Module : An agent that can speak naturally, human-like, and emotionally resonate.

• Personalized Agents : Agents with unique, relatable, and personalized conversations.

• Live Agents : Interactive agents that thrive on platforms like Twitter/X and YouTube.

• Automated trading agents : Capable of consistently executing profitable trades.

• DeFi- focused agency : optimizing yield strategies, lending, and liquidity configuration.

• Abstract Proxy : Enable seamless on-chain interactions through a user-friendly UI.

Just as humans are diverse and specialized, AI Agents will become equally diverse. The uniqueness of each agent will be closely tied to its underlying models, data, and infrastructure. However, the success of this ecosystem depends on a strong decentralized AI infrastructure.

The role of decentralized AI infrastructure

Decentralized infrastructure is critical to scaling AI agents in 2025; without it, the field risks bottlenecks in performance, transparency, and innovation.

Here’s why each part of decentralized AI infrastructure is important, and the projects currently being built to address these challenges:

Verifiability

Trust is the foundation of decentralized AI. As AI agents become more autonomous, we need systems that allow us to verify what is happening in the background. We need to know whether this agent is a real AI or disguised as a human; whether the output is accurate and generated by the claimed algorithm or model; whether the calculation is performed correctly and securely, and so on.

This also involves the Trusted Execution Environment (TEE), which ensures that the agent runs independently, securely, and cannot be manipulated. Without verifiability, there is no trust, and without trust, the ecosystem cannot scale.

Notable projects:

• ORA : Exploring infrastructure for secure AI, but token economics still needs to be improved;

• Hyperbolic : pioneered sampling proofs for verifying AI computations and reasoning;

• Phala Network : Known for its TEE infrastructure, which adds a layer of security to decentralized AI.

Paiement

For AI agents to operate autonomously in the real world, they need payment systems. Whether transacting with humans or other agents, these systems must handle everything from on-chain/off-chain to barter and accounting. Imagine agents that can independently manage finances, purchase computing resources, and even exchange services with other agents – this is the basis for commercial transactions between agents.

Notable protocols:

• Crossmint : AI payment tool to facilitate transactions;

• Nevermined : Supports commercial transactions and interactions between agents;

• Skyfire : Focuses on payments and accounting for agency operations.

Decentralized computing

The computational demands of AI are skyrocketing—doubling approximately every 100 days. Traditional cloud services such as AWS cannot keep up with this demand, both in terms of cost and accessibility. Decentralized computing networks allow anyone with spare resources to join the network, provide their computing power, and earn rewards.

This year, we’ve even seen the emergence of GPU-backed debt financing models, such as GAIB , helping data centers finance and scale their operations. This makes decentralized computing accessible to a wider audience.

Notable protocols:

• Aethir : Decentralized computing tailored for AI and Web3.

• io.net : Provides scalable computing solutions for AI workloads.

données

If AI is the brain, then data is the oxygen. The quality, reliability, and integrity of data directly impact the performance of AI models. However, acquiring and labeling high-quality data is expensive, and bad data leads to bad results.

Excitingly, there are platforms emerging that allow users to own and monetize their data. For example, vana allows contributors to tokenize their data and trade it in a data liquidity pool (DLP). Imagine choosing a TikTok DataDAO or a Reddit DataDAO to pool your contributions — a concept that empowers users while advancing AI.

Notable protocols:

• Cookie DAO : A reliable source of data metrics and insights;

• vana : Tokenize user data into liquidity pools that can be traded on decentralized markets;

• Masa : Partnering with Virtuals to build the largest decentralized AI data network to power dynamic and adaptive AI agents.

Model Creators and Marchéplaces

2025 will see an explosion of new AI agents, many of which will be powered by decentralized models. These models will be more advanced, incorporating human-like reasoning, memory, and even cost awareness.

Par exemple, Nous Research is working on a “starvation” mechanism that introduces an economic constraint to AI models. If the agent cannot afford the cost of reasoning, it effectively “dies,” teaching it to prioritize tasks more efficiently.

Notable projects:

• Nous Research : Introducing a “starvation” mechanism to teach AI resource management;

• Pond : Partners with Virtuals to provide tools for decentralized model creation and training;

• Bagel : Providing privacy-preserving infrastructure using FHE and TEE.

Distributed Training and Federated Learning

As AI models become larger and more complex, centralized training systems will no longer be able to meet the needs. Distributed training spreads the workload across multiple decentralized nodes, making the process faster and more efficient. En même temps, federated learning allows organizations to collaborate on training models without sharing raw data, thus addressing major privacy issues.

FLock.io is the Uber for AI. Flock connects AI engineers, model proposers, and data providers to create a marché where AI models can be trained, validated, and deployed in a secure and decentralized manner. It powers projects like Aimonica and many more interesting models.

Swarm Intelligence and Coordination Layer

As more specialized agents enter the ecosystem, seamless communication between them becomes critical. Swarm intelligence allows agents to work together as a team, pooling their capabilities to achieve a common goal. The coordination layer abstracts the complexity, making it easier for agents to collaborate.

Par exemple, Theoriq uses meta-agents to identify the agents best suited for a task and form “swarms” to achieve a goal. It also tracks reputation and contributions to ensure quality and accountability.

Notable projects:

• FXN : Creates a protocol for unified communications and commerce;

• Virtuals : Enables interaction and integration between agents;

• Theoriq : Develop agents and build advanced coordination tools for AI agents, including clustering and task delegation.

Why decentralized infrastructure matters

The next phase of AI agent development depends on infrastructure. Without verifiability, payment systems, scalable computing, and a robust data pipeline, the entire ecosystem risks stagnation. Decentralized infrastructure helps solve these problems by providing trust and transparency, scalability, collaboration, and empowerment.

Of course, there are several other narratives expected to develop in 2025, such as:

• Agent Metaverse/AI x Games: Projects like Realis and ARC Agents are fusing agents with games and immersive virtual worlds;

• On-chain and DeFi tooling: Protocols such as Almanak , Wayfinder , Axal , Cod 3x , Griffain , et Orbit are building essential tools for DeFi-powered brokers.

Résumer

2025 will be the year of the AI Agent, when we’ll see them rapidly move toward sentient AGI. These agents won’t just perform isolated tasks — they’ll trade autonomously, collaborate with other agents, and interact with humans in ways we can’t even imagine.

Imagine an agent analyzing market data, making trades, managing your finances, or coordinating with others to complete complex tasks. They will be seamlessly integrated into our lives, handling everything from on-chain DeFi operations to real-world interactions, with autonomy and intelligence far beyond what we see today.

The decentralized infrastructure (verifiable systems, payment tools, computational networks, and coordination layers) currently being built will make this future possible. For builders, investors, and enthusiasts, now is the time to dive in and shape that future.

2025 is not just a continuation, but also the dawn of the next era of AI Agents.

This article is sourced from the internet: AI Agents 2024 Review and Outlook: Where Have We Come From and Where Are We Going?

Related: Spotlight on Binance MVB Season 8 Project Alias: AI Agent Innovation Fusion of Crypto Technology

As intent-based transactions become the new form of internet entry, tech giants are doubling down on AI agents. As a leader in the AI industry, OpenAI classifies AI agents as the third layer of achieving AGI (artificial general intelligence), positions them as the Next Big Thing, and sets 2025 as a clear release schedule. Companies such as Microsoft, Facebook, and Google have launched AI agent programs, which not only demonstrate the potential applications of large language models (LLMs), but also reignite the entrepreneurial enthusiasm around LLMs. Behind the surge in interest from these tech giants is the emergence of a new form of internet entry, evolving from search engines to intent-based transactions—a market estimated to be over $10 trillion that will cover scenarios in daily life, work, social interactions, and…