From AI Agents to Swarms: Where Value Accrues in the AI Economy

Executive Summary

• AI frameworks like Eliza, ZerePy, and Dolion are driving the creation of modular, specialized AI agents that serve diverse functionalities across platforms.
• The next step in AI’s progression is swarms — interoperable ecosystems of agents collaborating toward complex goals. These swarms amplify individual agent capabilities, enabling tasks like website building and trading through collective intelligence.
• Similar to blockchain ecosystems, AI swarms face significant hurdles in achieving interoperability across frameworks like Virtuals, FXN, and UBC. Bridging solutions may emerge, but they will introduce complexities akin to cross-chain technologies in crypto.
• Frameworks Serve as foundational layers for creating AI agents and swarms. AI Agents Deliver specialized, user-facing value, driving engagement and utility. AI Swarms Unlock exponential efficiency by combining agents into task-focused ecosystems. Tokenomics: The introduction of tokens within frameworks, agents, and swarms will shape incentives, governance, and monetization.
• A framework’s success hinges on its ability to launch high-quality, impactful agents that attract developers and users. Network effects will drive consolidation around dominant frameworks as swarms form around them.
• AI swarms could revolutionize industries by enabling decentralized and efficient task execution. Fragmentation across frameworks may persist, creating silos and necessitating bridging solutions. Developers and investors must carefully navigate risks and opportunities to align with the most promising frameworks and swarms.
• The transition from standalone agents to swarms signals a transformative shift in AI. Key questions remain: Can interoperability be achieved? Which layer — frameworks, agents, or swarms — will capture the most value?

The Rise of Modular AI Frameworks

The AI ecosystem is transforming with the emergence of modular frameworks like Eliza, ZerePy, and Dolion. These frameworks are revolutionizing the creation and deployment of AI agents, offering powerful tools to design specialized agents tailored for various use cases, from social media management to knowledge work.

At the forefront of this innovation, Eliza stands out as a multi-agent simulation framework that enables the creation of intelligent, personality-driven agents capable of operating across multiple platforms. With features like long-term memory management, seamless integration with social platforms like Discord and Twitter, and support for advanced language models, Eliza has already powered notable agents such as @DegenSpartanAI and @MarcAIndreessen, showcasing its flexibility and potential.

Similarly, ZerePy, developed by the Zerebro team, offers a Python-based framework designed to bring AI interactivity to platforms like X (formerly Twitter). Its functionalities include automating tweets, replies, and social media engagement while leveraging the power of OpenAI and Anthropic LLMs. ZerePy’s vision is to enhance user control over AI-driven interactions, promising future support for memory, local LLMs, and expanded platform integration.

On the other hand, Dolion takes accessibility to the next level by introducing a no-code, one-click deployment framework. Built on the Dolos architecture, Dolion enables users to create cross-platform AI agents with ease. By simply defining an agent’s personality and selecting features, even non-technical users can deploy agents for tasks like posting on TikTok, replying on Telegram, or engaging in on-chain trading. Powered by the Primis protocol, Dolion lowers the entry barrier, unlocking possibilities for widespread AI adoption.

Star history for AI agents/frameworks on GitHub

These frameworks collectively represent the foundational layer of the AI ecosystem, empowering developers, brands, and individuals to create modular, specialized AI agents. However, the true potential of these agents lies not in their standalone capabilities but in their ability to collaborate, forming AI swarms — interconnected groups of agents working together to achieve complex goals.

As these frameworks evolve and gain traction, the next frontier in AI development will likely shift from building individual agents to creating ecosystems where these agents can seamlessly interact. This concept of AI swarms signals a paradigm shift in how we think about AI’s role in everyday life, much like how blockchain ecosystems evolved from standalone protocols to interconnected DeFi systems.

From AI Agents to AI Swarms

Source

The current wave of AI innovation has focused on the development of standalone AI agents — specialized entities designed to perform unique tasks independently. Frameworks like Eliza, ZerePy, and Dolion have enabled the creation of such agents, empowering users to deploy tools tailored to specific goals. For instance, Eliza’s AI16z demonstrates how AI agents can manage and engage with social media autonomously, while other agents might specialize in customer support, trading, or content generation. These agents embody the core promise of AI: highly customizable, task-oriented solutions that bring intelligence to diverse domains.

However, the next natural step in this evolution lies in AI swarms — collections of interoperable AI agents working collaboratively toward a shared objective. Unlike standalone agents, which operate in isolation, swarms represent a networked approach where individual agents contribute their specialized capabilities to a larger, coordinated effort. For example, building a website could involve a swarm of agents: one specializing in design, another in front-end development, and a third in back-end integration, all communicating seamlessly to deliver the final product.

This shift mirrors the progression of blockchain ecosystems. In the early days, Layer 1 (L1) blockchains like Ethereum laid the foundation for standalone applications. Over time, however, interconnected systems emerged in the form of DeFi protocols, where smart contracts from projects like Uniswap and Aave could interact. These protocols created tokenized economies, unlocking new possibilities for decentralized collaboration.

AI swarms, in a similar vein, have the potential to function as the AI economy of the future. Each agent within a swarm could represent a specialized service, tokenized for incentivization and governance. The frameworks enabling these swarms, like Eliza and ZerePy, serve as the equivalent of blockchain L1s, providing the infrastructure for interoperability.

Yet, this evolution brings challenges. As with blockchain ecosystems, achieving seamless communication between agents and swarms built on different frameworks will require robust standards and protocols. Without them, AI swarms could fragment into isolated ecosystems, much like today’s blockchains struggle with interoperability.

Still, the promise of AI swarms is compelling. They represent a future where modular, task-specific AI agents work together seamlessly, solving problems beyond the reach of standalone systems. By leveraging their combined intelligence, AI swarms could redefine how we approach everything from business automation to personal productivity, much like interconnected blockchain protocols revolutionized decentralized finance.

The Interoperability Challenge

Source

The transition from standalone AI agents to fully interoperable AI swarms is one of the most exciting prospects in artificial intelligence — but it comes with significant challenges. Frameworks like Virtuals, FXN, UBC, and Arc aim to facilitate the creation of modular, specialized AI agents. These frameworks are designed with the potential to serve as the infrastructure layer for AI swarms, where agents can interact and collaborate seamlessly. However, in their current state, these ecosystems are largely siloed.

Agents built on Virtuals, for instance, cannot inherently communicate with those on FXN without additional layers of abstraction or middleware. This lack of interoperability could become a major bottleneck for AI swarms, as complex tasks often require the cooperation of agents across frameworks — something we’ve yet to see achieved at scale.

The Blockchain Parallel

This challenge mirrors the interoperability issues blockchain ecosystems face, where competing Layer-1 chains like Ethereum, Solana, and Avalanche operate independently. Assets or applications on one chain cannot natively interact with those on another without using bridges or other external solutions. Even today, over a decade into blockchain’s development, interoperability remains a work in progress, with no universally adopted solution.

In the context of AI swarms, a similar problem arises:

• Framework-specific ecosystems: Virtuals, FXN, and other frameworks are developing unique architectures and protocols to serve as the “base layer” for AI agents. However, their lack of a shared communication protocol limits the ability of agents from different ecosystems to collaborate.
• The emerging dominance of specific frameworks: Just as Ethereum became the de facto choice for many DeFi projects, frameworks like Virtuals or FXN might establish dominance, creating centralized hubs for AI swarm development. But as history shows, this dominance often breeds fragmentation rather than unity.

Complexities in Bridging AI Frameworks

Bridging different AI swarm frameworks, such as Virtuals and FXN, presents several complex challenges. One key hurdle is data translation. Unlike blockchain bridges that primarily transfer tokens or data, AI swarm bridges must translate context-specific and task-oriented instructions between agents. This requires a deeper understanding of the data being exchanged. Another challenge lies in aligning memory and goals. Frameworks like Virtuals may manage agent memory differently than FXN, which could lead to conflicts in how agents align their long-term goals and contextual understanding. This difference in memory management can create inconsistencies in agent behavior.

The dynamic nature of AI agent interaction adds another layer of complexity. AI agents require dynamic decision-making and real-time communication, making interoperability more complex compared to deterministic systems like blockchains. This need for real-time interaction requires sophisticated communication protocols.

Differences in underlying architectures, such as Dolion’s no-code, token-based deployment model versus Eliza’s TypeScript-based modular system, can create compatibility issues. These architectural differences may necessitate the development of meta-frameworks or middleware layers to act as interpreters between the different systems.

The proprietary nature of some frameworks could further complicate interoperability. Open-source frameworks like ZerePy might prioritize openness, while others may enforce restrictions to maintain control over their ecosystems. These differences suggest that interoperability in AI swarms might require new meta-frameworks or middleware layers designed specifically to facilitate communication between frameworks.

Lessons from Blockchain Interoperability

Blockchain interoperability provides valuable insights for developing interoperable AI swarms. Just as blockchain ecosystems developed token standards like ERC-20 and ERC-721 to ensure compatibility, AI frameworks may need to adopt shared standards for agent communication, task delegation, and data formats.

Similar to blockchain bridges that facilitate asset movement between chains like Solana and Ethereum, we might see the emergence of AI swarm bridges. These middleware solutions would enable agents from different frameworks to interact, allowing agents from systems like Virtuals and FXN to share context, memory, and tasks. However, like their blockchain counterparts, these AI bridges would likely face trade-offs between security and efficiency.

Mirroring the consolidation seen in the blockchain space, where a few chains dominate, AI swarms might also consolidate around a handful of frameworks. While this could reduce fragmentation, it also carries the risk of creating centralized power dynamics.

Future Prospects

The path to interoperability is long and fraught with challenges. Just as blockchain interoperability remains a work in progress after a decade, AI swarms may also face years of experimentation before achieving seamless communication across frameworks. The frameworks that solve this problem — whether through standardized protocols, bridging layers, or a dominant ecosystem — will likely hold immense value.

Without interoperability, the risk is a fragmented future where swarms built on Virtuals, FXN, or other frameworks remain isolated, unable to unlock their full potential. But with it, we could see the emergence of a truly interconnected AI ecosystem, capable of executing complex, multi-agent tasks at scale.

This makes interoperability not just a technical challenge but also a massive economic opportunity — one that could shape the future of AI in much the same way Ethereum shaped DeFi.

Where Does Value Accrue in the AI Ecosystem?

In the AI ecosystem, value creation — and by extension, investment potential — is distributed across several layers. Each layer serves a unique role in building the broader AI landscape, from foundational frameworks to the user-facing applications of AI swarms.

Frameworks: The Foundational Layer

Frameworks like Virtuals, FXN, and UBC form the base layer of the AI stack. They enable the creation of AI agents by providing developers with tools to design, deploy, and manage autonomous agents. Their role is analogous to Layer-1 blockchains in crypto, such as Ethereum or Solana:

1. Infrastructure Providers: Frameworks handle core functionalities like memory management, data retrieval, and communication protocols for agents.
2. Ecosystem Builders: Just as Ethereum hosts DeFi apps, frameworks foster ecosystems of agents and swarms, each with distinct use cases.

For example, Virtuals focuses on enabling modular swarms through flexible architecture, while FXN is optimizing for cross-platform integrations, positioning themselves as key players for interoperability. ZerePy and Dolion, though currently focused on standalone agents, may evolve into frameworks supporting swarms.

Value Accrual: Frameworks that attract the most agents and swarm builders will capture value through fees, token burns (e.g., Dolion’s $Bully), and network effects. Developers gravitate toward frameworks that lower friction and provide powerful modular tools, making them pivotal for the ecosystem’s growth.

AI Agents: User-Facing, Specialized Entities

AI agents represent the frontlines of value delivery, acting as the interface between frameworks and end users. These agents are designed to perform highly specialized tasks, such as: Automating Investment strategies (e.g., AI16z), managing personal and professional social media accounts (e.g., LUNA), providing unique virtual personalities for entertainment or productivity (ZEREBRO), etc.

Agents built on frameworks like Eliza or ZerePy have showcased early successes, offering tangible user benefits and proving that standalone agents can generate substantial engagement. However, they face scalability challenges when the tasks exceed the scope of a single agent’s capacity, paving the way for AI swarms.

Value Accrual: Individual agents could capture value through subscription models, freemium tiers, or utility tokens tied to their services. The uniqueness and utility of agents will determine their stickiness in the ecosystem, but they are inherently less scalable compared to the frameworks or swarms they belong to.

AI Swarms: Collaborative Ecosystems of Agents

AI swarms take the capabilities of standalone agents and amplify them through collaboration. These swarms, composed of interoperable agents, are designed to tackle multi-faceted, complex tasks that no single agent could handle efficiently.

For example:

• A website-building swarm could combine agents specializing in design, SEO, content generation, and analytics, all coordinating seamlessly.
• A trading swarm might integrate strategies from multiple specialized agents for optimal portfolio management.

In blockchain terms, swarms are comparable to ecosystems of protocols built atop L1 chains — modular and purpose-driven, yet collectively more powerful than individual agents. Swarms also introduce network effects: as more agents join, the swarm’s utility grows, attracting more users and developers.

Value Accrual: Swarms are likely to monetize through task-based fees, revenue-sharing models, or their own native tokens. Tokenized swarms could represent investment opportunities for users, allowing them to stake in or govern the swarm’s operations.

Tokenomics

Tokens play a critical role in aligning incentives across the AI ecosystem. Each layer has its own token model, creating diverse investment opportunities:

1. Framework Tokens: Frameworks like Dolion already integrate token mechanics (e.g., $Bully for agent deployment). Future frameworks might adopt models similar to Ethereum, where tokens are required for computation, agent training, or memory upgrades.
2. Agent-Specific Tokens: Unique agents might launch their own tokens to monetize specific capabilities. For example, users could stake tokens to unlock premium features or governance rights.
3. Swarm Tokens: The most exciting tokenomics might emerge from swarms, where tokens govern swarm behavior, incentivize agent collaboration, and share in the revenue generated by the swarm. These tokens would operate like DAO governance tokens in crypto, granting holders decision-making power.

The success of tokens will depend on network effects, utility, and scarcity. Framework tokens, due to their foundational role, might represent the safest bet, while swarm tokens offer the most upside but carry higher risk. Agent-specific tokens, though smaller in scope, could thrive if tied to high-value, niche applications.

Convergence and Competition

A crucial aspect to consider is how these different layers will interact and potentially compete over time. One key question is whether frameworks like Virtuals will aim for dominance by developing their own agents and swarms, or if they will adopt a more agnostic approach, allowing third-party developers to build on their platforms. Another important consideration is whether agent creators will consolidate under a few popular swarms, or if a more fragmented landscape of independent ecosystems will persist. The question of incentive alignment through tokenomics arises: will it be possible to align incentives across all layers of the ecosystem, or will competing standards emerge, similar to the early days of blockchain technology?

These dynamics will determine where value truly accrues in the AI ecosystem and which players — frameworks, agents, or swarms — capture the lion’s share of the market. This layered view highlights the massive potential for innovation and investment in the AI ecosystem, while also emphasizing the interdependence of its components. As frameworks, agents, and swarms evolve, their ability to interoperate and create value collaboratively will define the future of AI-driven economies.

Frameworks as the Battleground

The frameworks powering AI agents and swarms — such as Virtuals, FXN, Eliza, ZerePy, and Dolion — are at the heart of this burgeoning ecosystem. Their role is pivotal: frameworks establish the rules, infrastructure, and development environment for agents and swarms to thrive. However, the competition among frameworks is shaping up to be one of the most critical battlegrounds in the AI space.

The Importance of Initial Agents

The success of any framework is heavily tied to the quality and utility of the first agents it powers. Just as Ethereum’s early adoption by decentralized applications (dApps) like Uniswap and Compound solidified its dominance, the agents built on frameworks like Eliza and ZerePy act as proof of concept for the framework’s capabilities.

Frameworks like Eliza showcase their potential through successful agents such as DegenSpartanAI and AI16z, which demonstrate the practical applications of multi-agent architectures. Similarly, Dolion’s no-code approach could attract non-technical creators, rapidly expanding its pool of agents and potential use cases. Virtuals and FXN, with their focus on swarm-enabled architectures, are strategically positioned to lead the next phase of innovation, where agents transition into collaborative ecosystems.

Early success stories create network effects, attracting more developers to build on the framework. Developers want to work within ecosystems where: Tools are robust and easy to use, there’s an existing audience or user base for their agents, the framework has proven its ability to scale. A thriving ecosystem of agents incentivizes further development, creating a feedback loop that entrenches the framework’s position.

Network Effects: The Swarm Advantage

Frameworks that enable the creation of AI swarms gain a critical competitive edge. Succeeding in this domain means frameworks can go beyond individual agents to facilitate modular, interoperable ecosystems For example, a swarm for automated trading could incorporate agents for portfolio analysis, market sentiment tracking, and execution strategies, all working seamlessly within the same framework. These swarms become more attractive as their scale and diversity of capabilities grow, driving adoption from both developers and end users.

Virtuals and FXN, with their early focus on enabling swarms, are laying the groundwork for these complex ecosystems. If their frameworks prove successful, the swarms they host will attract further development, solidifying their dominance. Dominant frameworks will capture network effects as developers increasingly align with them, leading to a “winner-takes-most” dynamic. Swarms built within these frameworks could also enforce lock-in effects, making it harder for developers or users to migrate to competing frameworks.

Incentives for Developers and Users

For frameworks to achieve dominance in the AI ecosystem, they must offer compelling incentives for both developers and end users.

To attract top developer talent, frameworks need to prioritize several key areas. Ease of use is crucial, with tools like Dolion’s no-code deployment lowering entry barriers and enabling broader participation. Revenue opportunities, such as enabling tokenized agents or offering profit-sharing models like those seen in Virtuals or Dolion’s $Bully burn mechanism, are also essential for incentivizing development. Developers will favor frameworks that ensure interoperability and scalability, allowing agents to integrate into swarms or interact across different platforms.

From the end-user perspective, framework choice often hinges on agent utility. Users will naturally gravitate toward agents that provide proven value, which indirectly benefits the frameworks hosting them. Ease of access is another critical factor, with frameworks that enable seamless swarm interactions or interoperable agents across platforms like Virtuals and FXN enhancing user experience and loyalty. Cost efficiency, particularly through token-based ecosystems that can subsidize user costs, also adds significant appeal.

The Stakes of the Framework Race

The battle for dominance among frameworks is not just about capturing developer mindshare — it’s about defining the rules of engagement for the AI economy. Frameworks that attract the most swarms and agents will control key data flows and revenue streams, much like dominant platforms in Web2 and L1s in crypto. The standards set by winning frameworks will influence interoperability, tokenomics, and even the broader trajectory of AI swarm development.

However, this dominance will require continuous innovation. As competing frameworks experiment with new architectures, token models, and incentives, the race for dominance will remain fluid, with room for disruptive players to emerge. In the end, the framework that achieves early adoption through superior tools incentivizes agent creation and swarm formation, and fosters network effects will likely define the future of the AI ecosystem. For developers, users, and investors, aligning with the right framework early could be the key to capturing long-term value in this rapidly evolving space.

Implications for the Future AI Economy

As AI agents evolve into swarms, the broader implications for industries, developers, and investors are profound. These ecosystems could become the backbone of the next-generation AI economy, transforming workflows, creating new markets, and reshaping how industries operate.

AI Swarms as Industry Disruptors

AI swarms represent a paradigm shift from standalone agents to coordinated networks of intelligent entities. By enabling agents to collaborate and specialize, swarms can tackle complex, multi-faceted tasks with greater efficiency.

Examples of Industry Impact:

1. Website Building: Imagine a swarm where agents handle user interface design, SEO optimization, and personalized content generation in real-time. A single command could yield a fully optimized, ready-to-launch website. Frameworks like Virtuals or FXN could dominate this space, providing modular agents that specialize in each aspect of website creation.
2. Trading and Finance: Trading-focused swarms could combine agents for market analysis, sentiment tracking, execution, and risk management. Virtuals’ modular architecture could lead to specialized trading swarms that adapt dynamically to market conditions.
3. Customer Support: AI swarms could integrate agents trained in language processing, sentiment analysis, and problem-solving to deliver superior customer service across multiple platforms simultaneously. Frameworks like Eliza, with advanced multi-platform support, are well-positioned to lead in this domain.

Swarms will likely redefine productivity by automating highly specialized and collaborative tasks, reducing reliance on human intervention and dramatically increasing efficiency.

Convergence vs. Fragmentation

A critical question for the AI economy is whether swarms will converge around a few dominant frameworks or whether a fragmented landscape will persist.

As in blockchain, network effects could lead to a few dominant frameworks (e.g., Virtuals or FXN) capturing the majority of developers and swarms. These dominant frameworks could establish standards for interoperability, reducing friction for developers and users. Convergence would create centralized hubs for swarm development, simplifying the ecosystem but potentially introducing monopolistic risks.

Alternatively, fragmentation could mirror the proliferation of Layer 1 blockchains in crypto, where niche frameworks cater to specific use cases or communities. For example, Dolion’s no-code deployment may attract creators focused on social media agents, while frameworks like ZerePy cater to technically skilled developers needing deep customization. Fragmentation would foster innovation but could exacerbate interoperability challenges, making it harder for agents and swarms to interact seamlessly.

Which is Likely? The trajectory may initially trend toward fragmentation as competing frameworks innovate, followed by eventual convergence around a few dominant platforms that prove their scalability and reliability.

Opportunities for Investors and Developers

The AI swarm economy, being in its early stages, presents both opportunities and risks for those who participate early. From an investor’s perspective, several avenues exist. Investing in dominant frameworks, such as Virtuals or FXN, could prove valuable in the long term if these frameworks capture a substantial portion of agent and swarm activity. Furthermore, frameworks with well-designed tokenomics, like $Bully for Dolion, may offer additional benefits through staking, governance mechanisms, or token burns. Investors can also consider directly investing in high-performing AI agents and swarms, particularly those focused on specific tasks like trading. These agents can generate revenue directly from users or through tokenized business models. Tokenized swarms themselves can function as diversified investment vehicles, similar to decentralized autonomous organizations (DAOs).

For developers, the nascent nature of the AI swarm economy offers a significant early-mover advantage. Building agents on emerging frameworks provides an opportunity to establish a strong presence within rapidly growing ecosystems. Successful agents have the potential to become fundamental components of future swarms. Developers who specialize in optimizing swarm performance, particularly in areas like interoperability, could find themselves in high demand as swarms continue to scale. However, developers also face challenges, including navigating a landscape of fragmented frameworks and overcoming potentially high technical barriers. Frameworks like Dolion, which offer no-code options, may help reduce friction for creators who lack extensive technical expertise.

Risks to Watch

Several challenges could hinder the widespread adoption and effectiveness of swarms. First, an over-reliance on dominant frameworks could lead to scalability bottlenecks, mirroring Ethereum’s congestion issues and potentially incentivizing fragmentation. Second, interoperability failures, in the absence of robust bridging solutions, could result in siloed swarms operating on different frameworks, limiting the overall potential of the ecosystem. Third, the risk of tokenomics saturation exists, with the potential for over-proliferation of tokenized models leading to user fatigue or value dilution. Therefore, careful design of token incentives will be crucial. Finally, as swarms take on more sensitive tasks, such as financial trading or data analysis, increased regulatory scrutiny could impact the pace of innovation.

The future of the AI economy is one of immense potential and complexity. As swarms reshape industries and frameworks vie for dominance, the stakes for developers, users, and investors will only grow. The key to navigating this evolving landscape lies in understanding where value accrues — and aligning with the ecosystems most likely to succeed.

The Road Ahead for AI Swarms

The transition from standalone AI agents to interoperable swarms represents the next major leap in artificial intelligence. Frameworks like Virtuals, FXN, and Dolion are laying the groundwork for this evolution, enabling the creation of agents that are increasingly specialized and capable. However, the true potential lies in their ability to work together within AI swarms, reshaping how industries operate, creating new markets, and driving efficiency across domains.

As we stand at the beginning of this transformative period, several key questions remain open. A critical question is whether interoperability can be effectively achieved. Will frameworks and developers prioritize creating seamless collaboration between different swarms, or will fragmentation continue to be a significant obstacle? Another crucial question concerns value capture within the ecosystem. Will the underlying frameworks become dominant as the essential infrastructure, or will individual swarms and agents emerge as the primary drivers of value? The answers to these questions will ultimately shape the evolution of the AI economy and determine where the most promising opportunities for innovation and investment lie.

For developers, investors, and enthusiasts, this is a crucial moment to engage with the evolving AI swarm ecosystem. Whether by building on promising frameworks, contributing to interoperability solutions, or investing in emerging agents and swarms, there is an opportunity to help shape this next frontier in AI. Stay informed, stay curious, and prepare to navigate a world where AI swarms become as ubiquitous as blockchain ecosystems are today. The future is collaborative, modular, and rapidly approaching.