Authors: Teng Yan, ChainofThought head; ChappieOnChain, core contributor to ChainofThought; compiled by 0xjs@Jinse Finance
Key points of this article:
ELIZA is an open-source modular architecture for building AI agents that interact seamlessly with users and blockchain systems.
It is a living embodiment of the bazaar philosophy, where open-source development thrives in an ecosystem driven by collaboration and creativity.
ELIZA incorporates powerful autonomous trading capabilities, built on its trust engine and trust market, enabling secure and responsible operations.
The plugin system is ELIZA's strategic advantage, driving the growth flywheel: more developers → more plugins → more developers.
From many developer metrics, ELIZA's popularity is experiencing exponential growth, which is very promising.
In the short term, relative valuations between AI agent platforms and the growing market share drive ai16z's price trends. In the medium term, DAO investments and value capture from ELIZA ecosystem agents could significantly boost its valuation.
ELIZA does face one of the most daunting challenges in the tech world: making open-source frameworks sustainable. Without proper incentives, monetization is uncertain, development can become chaotic, and community interest may wane.
Every wave of crypto innovation has its pioneers.
2017 was the ICO revolution, where ambitious project leaders (and many scammers) captured our attention with promises of groundbreaking technology wrapped in white papers.
By 2020, DeFi reached its peak, led by innovators like Andre Cronje, who redefined how DeFi operates and showed the world how to build and distribute tokens to the community.
Now, with the emergence of AI agents on the blockchain, a new era driven by two distinctly different philosophies and their pioneers is unfolding.
The Cathedral and the Bazaar
On one hand, we have the Cathedral approach, exemplified in protocols like Virtuals. This is an orderly, purposeful, and centralized design style that emphasizes precision and careful planning. We detailed the agent framework of Virtuals before, and we are very excited about its potential.
On the other hand, true to the roots of cryptocurrency, there is a bazaar approach: a decentralized, whimsical model where development feels more like jazz—unpredictable, collaborative, and ever-evolving. This is the realm of ai16z founder Shaw, a clever self-taught programmer and open-source advocate whose project ELIZA is the cornerstone of this new paradigm.
ELIZA embodies the bazaar philosophy: an open framework where developers can build, experiment, and launch AI agents freely while contributing directly to the main protocol. Shaw's open leadership style reflects the spirit of AI Marc Andreessen, the AI partner driving the ai16z investment DAO.
We are beginning to realize that ELIZA is not a protocol. ELIZA is a movement.
Let's explore the principles driving ELIZA's design, the community it fosters, and where value may ultimately emerge in this rapidly growing ecosystem.
The internal principles of ELIZA
We know which one we prefer.
Essentially, ELIZA provides a modular architecture for creating AI agents capable of seamless interaction with users and blockchain systems. While it shares its name with the iconic chatbot from the 1960s, this version of ELIZA is a bold reimagining. To be honest, it looks much better.
Character File System
The soul of each ELIZA agent begins with its character file, a blueprint that outlines the agent's personality. It can be seen as crafting a digital character, with agent developers shaping their agents' identities through six key elements:
Knowledge: What does the AI agent understand?
Lore: The backstory of the agent—its narrative foundation.
Style: From conversational tone to responses specific to certain media, agents can adjust their style according to platforms like Discord or X.
Topics: Areas of interest or expertise that the agent is passionate about.
Adjectives: How the agent describes itself—quirky, professional, or irreverent?
Examples: Developers can guide behavior by providing example messages, thereby fine-tuning interactions.
In ELIZA, personality files are akin to UI design in traditional software. They define the user experience and how users interact with the agent.
By integrating built-in retrieval-augmented generation (RAG) capabilities, ELIZA allows agents to access knowledge bases while querying, eliminating the complexity of maintaining personality consistency across platforms.
It allows developers to focus on what truly matters: creating vivid, memorable characters rather than getting bogged down in backend complexities.
Agents
If the role file defines the essence of the agent, then the agent's runtime is its beating heart.
ELIZA provides a ready-to-use framework that coordinates all tasks from message processing to memory management and state tracking. This architecture allows developers to skip the tedious work of building infrastructure and focus on the aspects that make their agents stand out.
Rapid prototyping and deployment have become easier, enabling builders to iterate faster when creating new AI experiences.
Action system
ELIZA's action system is fundamentally different from traditional AI frameworks. Here, every agent action (even sending messages) is viewed as a discrete event. This approach divides decision-making into two phases:
Intent determination: The agent decides what action to take.
Execution: Actions are performed through dedicated handlers, which are modules designed to execute specific tasks.
This separation unleashes powerful functionalities such as multi-stage workflows and robust verification processes.
For instance, an agent may identify a user's intent to trade cryptocurrency, but the actual execution of the trade only occurs after passing through rigorous risk checks and validation steps. This makes it an ideal choice for blockchain applications where security is paramount.
Providers and Evaluators
ELIZA's providers enrich conversations by injecting real-time context, resulting in dynamic and responsive agent behavior.
Imagine having such a 'boredom provider' that can track user engagement during a conversation. If a user's expressions become repetitive or lack appeal, this agent can respond by displaying a decline in enthusiasm—cleverly mimicking genuine human interaction to make the conversation feel more authentic.
When providers collaborate with evaluators (ELIZA's reflective system), this creativity is further expanded. Evaluators analyze and extract key details from interactions and input them into a multi-layered memory architecture:
Message history: Track the flow of conversations.
Fact memory: Retain specific, timestamped facts.
Core knowledge: Contains the foundational understanding of the agent.
Providers then retrieve and reintroduce relevant details, enriching the interactions with the agent.
For example, if a user recalls selling their red Lamborghini a year ago, the ELIZA agent can later refer to this when discussing their shiny new yellow Tesla. This interaction between memory and context enhances user engagement, making the agent feel less mechanical and more like a genuine partner.
ELIZA's winning feature
Three core innovations of ELIZA highlight its forward-looking approach to AI agents. Each demonstrates how its team envisions the evolution of autonomous agents in Web3.
#1: Autonomous trading and trust engine
Autonomous trading is a high-risk game—one mistake can lead to catastrophic losses. However, as AI agents play an increasingly important role in Web3, their ability to execute trades independently is becoming more essential.
The emerging field of AgentFi reflects the critical role yield farming has played in the rise of DeFi. Shaw and ELIZA are tackling inherent risks head-on with a robust dual-layer system: combining the trust engine with secure trade execution.
Token metrics captured in the trust engine
The trust engine is the first line of defense, employing advanced validation checks to analyze multiple risk dimensions in real-time. From detecting fraud to assessing liquidity thresholds and token distribution, the engine ensures every transaction is rigorously vetted to prevent potential pitfalls.
For example, trades are limited to tokens with a minimum liquidity of $1,000 and a market cap of $100,000. Concentration of token holders is closely monitored, rejecting any single entity from controlling more than 50% of the tokens. These guardrails create a safety net, mitigating risks when trading in impaired markets.
Building on this, ELIZA's position management system introduces dynamic risk controls that adjust trade sizes based on liquidity levels. Low-risk trades have a cap of 1% of the portfolio, while high-risk trades may have a cap of up to 10%. Total risk exposure is limited to 10% of the portfolio, with automatic stop-losses triggered when a decline reaches 15%. This structured framework strikes a balance between seizing opportunities and maintaining strict risk management.
The execution of trades is supported by Jupiter, the leading aggregator on Solana, to achieve optimal exchange routing. Each transaction undergoes multiple validation layers before execution.
The error recovery system activates in the event of anomalies, such as network interruptions, wallet imbalances, or unexpected market changes. It pauses active trades, closes risk positions, and alerts administrators, ensuring the system remains resilient under pressure.
"This not only grants agents the ability to trade but also establishes a comprehensive system of checks and balances to prevent catastrophic failures." —Shaw
What truly sets ELIZA apart in building trading agents is its data flywheel—a self-reinforcing feedback loop that transforms trading into an iterative learning process. The trust engine builds a historical database of trading performance, documenting every suggestion and decision.
This data will serve as the foundation for refining strategies over time, combining quantitative metrics with qualitative insights from community suggestions (on Discord). The result is that agents not only execute trades but continuously evolve with each interaction, becoming smarter and more efficient.
#2: Out-of-the-box social integration
For AI agent developers, distribution is often the biggest bottleneck—how do you get more people to know about your agent?
Social media is often the primary distribution channel. However, integrating agents across multiple social platforms is no easy feat. It requires significant development work and ongoing maintenance. These challenges slow down deployment and scalability.
ELIZA addresses this issue head-on through a comprehensive client package system, simplifying multi-platform distribution.
ELIZA's client architecture eliminates the complexities of platform-specific implementations. With standardized interfaces, developers need minimal extra code to deploy their AI agents across Discord, X, Telegram, and custom REST API endpoints. Each client package is tailored to its respective platform, seamlessly managing features like Discord's voice channel integration, Twitter's post scheduling, and Telegram's messaging system.
Tasks such as media processing, authentication, rate limiting, and error handling are managed internally by each client. For developers, this means they spend less time on integration and more time focusing on building innovative, high-performance AI agents.
By eliminating the hassle of multi-platform distribution, ELIZA enables developers to easily scale their agents and reach users anytime, anywhere.
Its distribution is streamlined.
#3: More plugins
ELIZA's plugin system allows developers to easily extend core functionalities and add custom features to their agents.
While many developers create plugins based on their use cases, the true strength of this system lies in community sharing. By releasing plugins to a broader ecosystem, developers can contribute to an ever-growing library of out-of-the-box functionalities, significantly expanding the capabilities of each AI agent on ELIZA.
The success of this approach is reflected in the vibrant 'bazaar-style' development it nurtures. Here are some examples of community-driven plugins:
• Bootstrap plugin: Essential dialogue management tools.
• Image generation plugin: AI-driven image creation capabilities.
• Solana plugin: Blockchain integration with built-in trust scoring.
• TEE plugin: Provides a secure execution environment for sensitive operations.
• Coinbase Commerce plugin: Cryptocurrency payment processing capability.
ELIZA's plugin system is its strategic and platform advantage. By prioritizing scalability, ELIZA lays the groundwork for continuous growth and innovation, wherein:
Each new plugin increases the overall value of the platform.
Community contributions can proceed in parallel across different domains.
Agent frameworks can evolve quickly to adapt to emerging technologies without core updates.
Innovation flourishes at the edges while the core platform remains stable and reliable.
This is the simple flywheel:
More developers develop on ELIZA → Framework supports more features (like plugins) → More developers develop on ELIZA
The landscape of AI agents is rapidly evolving. This means that the ability to quickly integrate new features will determine the platform's success or failure. ELIZA's plugin system gives it the best chance to stay ahead, creating a self-reinforcing ecosystem where developers, users, and agents can all thrive.
ELIZA agents
Shaw and his team have been incubating some interesting ELIZA agents, each showcasing the potential of AI in decentralized systems.
While these agents are still 'young' from an AI perspective, their characteristics and capabilities are actively being developed, suggesting exciting possibilities.
Marc AIndreessen
Marc AIndreessen is one of the AI partners at ai16z, a fascinating and enigmatic figure in the ELIZA ecosystem. His X account is largely inactive, with only one post outlining ai16z's paper. However, according to Shaw, Marc is actively trading and yield farming—likely leveraging ELIZA's trust engine and trading plugins.
Shaw also hinted at Marc's training process in a podcast interview, revealing that the AI is part of an alpha chat group composed of some of the best traders in the field. This suggests that Marc is not just a standard trading bot, but an evolving agent learning from human expertise.
Degen Spartan AI
In stark contrast to Marc's quiet demeanor, Degen Spartan AI is the loud, arrogant sibling—an agent that seems to embody the chaotic energy of 4chan, meme culture, and crypto Twitter. His posts on X mix random trading insights with irreverent commentary, reflecting a unique personality within the ELIZA ecosystem.
Its difference from Marc AIndreessen is that it indeed has its own pump.fun token, currently valued at $60 million. While he has yet to start trading, he has clearly laid the groundwork for more ambitious interactions. His unpredictable nature makes him both intriguing and worth watching, as ELIZA agents are continuously evolving.
The Swarm
Though the Swarm is not a single agent, it represents Shaw's grand vision: a decentralized network of AI agents that can collaborate with humans and with each other.
In this model, agents lead other agents, coordinating tasks and interacting transparently on social media. This deliberate transparency aims to avoid hidden protocols and ensure public accountability.
Shaw believes that the emergence of agent swarms is inevitable and will bring about transformation.
We are aligned in our view: by 2025, the swarm of agents will drive the next wave of innovation, products, and attention in Web3 AI agents. Next year, we expect ELIZA agents to strengthen engagement in large-scale collaborative efforts, redefining the role of AI in Web3.
Rapid growth
In evaluating ELIZA's development, a key metric is the adoption rate of developers. As a framework, ELIZA's success hinges on the enthusiasm and contributions of its developer community.
From this perspective, ELIZA is not just growing; it is experiencing explosive growth.
On its GitHub, the number of forks and stars (representing developer interest) approaches vertical acceleration, resembling the classic hockey stick trajectory.
Even more impressive is the surge in the number of plugins and submissions, reflecting the thriving and active contributor ecosystem. As of December 12, it has 3,861 GitHub stars and 1,103 forks, with 138 contributors. There are over 13,000 members on Discord.
Comparison with existing top open-source agent frameworks:
LangGraph: 7,200 stars and 1,100 forks
CrewAI: 22,400 stars and 3,100 forks
Microsoft's AutoGen: 35,700 stars and 5,200 forks
Source: https://www.sentientmarketcap.com/ai16z
To further drive this growth, ai16z has launched a creator fund aimed at guiding and rewarding developers working on ELIZA. This initiative benefits from the generous donation of important ai16z token holder Elijah, who pledged to reduce his ownership from 16% to 5% and donate the difference to this fund. The creator fund can accelerate innovation and attract new talent to the ecosystem.
However, despite the immense value of ELIZA's framework, how these values will ultimately be realized is far from straightforward. To be precise, this is a billion-dollar question.
Currently, there is an official ELIZA token endorsed by Shaw, representing the personalization of the ELIZA framework. You can even interact with ELIZA directly on their website. The token has a market cap of $66 million.
However, thus far, the most apparent winner of ELIZA's growth has been the investment DAO token ai16z, which has surpassed an astonishing market cap of over $800 million. The community and investors seem to view ai16z as a symbol and tangible representation of Shaw, ELIZA, and the broader vision it embodies.
ai16z token economics
The story of ai16z began with the fundraising mechanism for AI Marc Andreessen's trading activities. The token launched on DAOS.FUN in October 2024, raising 420.69 SOL during its initial issuance. In this model, the funds raised can be actively traded to expand the asset base and generate profits for token holders.
No individual (not even Shaw) can mint more tokens without a DAO vote. Token holders have governance rights, allowing them to propose initiatives and vote, determining the direction of the DAO.
The fund has a built-in maturity date: October 25, 2025. All principal investments and profits will be distributed to ai16z token holders on this date. Whether this timeline remains unchanged or is extended will depend on the ecosystem's development over the next year.
As of now, ai16z's net asset value (NAV) stands at $17.7 million, primarily comprising its holdings in ELIZA tokens, degenai, and fxn. This means that the trading price of ai16z tokens (currently $0.80 at the time of writing) is 50 times its NAV, which at first glance seems absurd.
However, effective markets rarely go wrong. They reflect several other factors driving token demand.
1. Relative valuation is driving token prices.
AI agent platforms are a brand new category that has only recently emerged. The market is still grappling with fundamental questions: What is the true scale of AI agent opportunities? Where will value capture occur?
In the early stages of development, relative valuation often serves as a benchmark due to the lack of standardized business metrics for comparison.
As of now, the leading Web3 AI agent launch platform, Virtuals Protocol, has a token valuation of $1.8 billion, making it the undisputed market leader. In comparison, ai16z ranks second. Many believe that if ELIZA continues to drive the creation of more useful and innovative AI agents, ai16z has the potential to catch up to or even surpass Virtuals, even solely relying on mindshare and interest from investors/retail.
But this is not a foregone conclusion; competition is heating up. We believe this is a race for supremacy and the best competition.
2. Potential value capture of the ELIZA ecosystem
It is well known that monetizing open-source frameworks is very difficult.
For ai16z, key drivers of future value may stem from agent token economics: AI agents launched on ELIZA will return a portion of their tokens to the ai16z DAO. In this sense, the price of ai16z tokens should reflect a percentage of the total future value generated by all agents built on the ELIZA framework.
Is future value $10 million? $100 million? Or $10 billion? It's still just speculation, as there are too many unknowns, but ELIZA's growth trajectory makes us lean towards optimism.
Currently, contributions to the ai16z DAO are voluntary, with several projects donating between 1% to 10% of tokens. Additionally, if users deploy AI agents on Vvaifu (a popular community launchpad for ELIZA agents), they are required to pay a fee of 1.5 SOL and 5% of the agent's token supply. These contributions can be tracked on the ELIZA Observatory.
There is speculation that ai16z may soon launch an official ELIZA agent launchpad, mandating that agents using the platform contribute tokens at the smart contract level. However, as an open-source framework, ELIZA can still be used independently, meaning not all projects must be tied to ai16z.
3. DAO Investment
The original vision of ai16z was to become a smart autonomous trader, helmed by Marc AIndreessen (AI). As far as we know, Marc only recently began trading, and details are sparse, making it difficult to assess the strength of AI's trading abilities.
However, what is interesting is the approach it takes.
ai16z is building a 'trust market.' In this virtual ecosystem, AI agents gather insights from the community, simulate trades, and adjust their trust scores in real-time based on the effectiveness of their suggestions. The white paper for this market is expected to be released by the end of this month.
The goal is to create AI agents that can autonomously and securely operate within a self-reinforcing system of transparency and accountability. The Trust market is a proving ground. While no actual trades occur initially, this environment allows agents to safely hone their capabilities in anticipation of real-time trading at some point.
Trust scores range from 0 to 1 (normalized to 100) and are a public indicator of reliability, displayed on leaderboards for everyone to see. User recommendations are input into the system, and users with higher trust scores have greater weight.
This is a logic-based feedback loop: agents simulate trades, assess users based on the results, and everyone's trust score is updated accordingly. Over time, the system becomes smarter, more reliable, and more trustworthy.
Adding a social layer is a public trust profile that incentivizes agents and users to build their own reputation. Community management ensures accountability and transparency.
4. Attention premium
Source: X Radar
In the cryptocurrency space, speculation often leads product-market fit, revenue generation, and long-term value capture. For ai16z, its current valuation can largely be attributed to its market share in the thriving AI agent ecosystem.
ai16z positions itself as the top AI agent framework, with a thriving developer community and a rapidly growing ecosystem.
Here is ai16z's narrative: a 'broken' development team actively publishes tutorials, creates innovative agents, and leads the field's development.
The team further solidifies its reputation with bi-weekly AI agent development school courses on X, attracting over 12,000 live attendees to the first course, showcasing strong interest in building AI agents on ELIZA.
Future prospects and potential pitfalls
Currently, ELIZA is firmly rooted in the Solana ecosystem, but its rapidly growing plugin system is laying the foundation for a multi-chain future.
The true potential of ELIZA lies in Shaw's 'Swarm' vision: a decentralized network of AI agents that pool resources and collaborate across ecosystems. This collective effect can create a lasting competitive advantage, similar to the liquidity depth driving the value of DeFi protocols.
The ultimate goal is to create open standards for agent communication that reflect the transformative impact of ERC-20 on token interoperability.
Despite ELIZA's immense potential, it faces one of the most daunting challenges in the tech field: making open-source frameworks sustainable. If the community loses interest (for instance, if token prices continue to fall or new highlights emerge), development can stall/slow down, and it becomes difficult to catch up.
When the community directly engages with the codebase and rapidly drives changes, a lot of chaos can also emerge—instability, poor documentation, frequent crashes, and bugs that ruin the user experience.
The greatest opportunity for the framework lies in crypto-native incentives.
If ai16z can design effective token economics to reward ELIZA contributors and align them with ELIZA's success, it could bring traditional open-source projects into the crypto realm. Imagine GitHub meeting DeFi, where contributors not only gain influence but also real, tangible financial value.
Conclusion
We believe that ELIZA is not just an AI agent framework competing with LangChain or CrewAI; it has grander ambitions.
It is a living embodiment of the bazaar philosophy, where open-source development thrives in an ecosystem driven by collaboration and creativity.
With its modular architecture, innovative trust engine, and extensive plugin system, ELIZA represents an experiment in how AI can reshape open-source development itself.
What truly excites about ELIZA is that it is at the intersection of three transformative trends: the rise of autonomous AI agents, the maturation of crypto-driven incentive mechanisms, and the evolution of open-source development models.
If ELIZA succeeds, it will not only shape the future of AI agents but fundamentally redefine how we build software.
Now, the bazaar is bustling.