January 15, 2025
If ELIZA succeeds, it could fundamentally redefine how we build software.
Every wave of crypto innovation has its pioneers.
In 2017, it was the ICO revolution, where ambitious project leaders (and many grifters) captivated us with promises of groundbreaking technologies packaged within whitepapers.
By 2020, DeFi had its moment, led by innovators like Andre Cronje, who redefined how decentralised finance could function and showed the world how to build and distribute tokens to communities.
Now, as AI agents emerge on the blockchain, a new era is unfolding, driven by two distinct philosophies—and their trailblazers.
On one side, we have the cathedral approach, embodied by protocols like Virtuals. It’s a methodical, intentional, centralised style of design, emphasising precision and careful planning. We wrote in detail about Virtuals’ agentic framework earlier and we’re genuinely excited about its potential.
On the other side, true to crypto’s roots, there’s the bazaar approach: a decentralised, freewheeling model where development feels more like jazz—unpredictable, collaborative, and always evolving. This is the domain of Shaw, a smart, self-taught programmer and open-source advocate whose project, ELIZA, is the cornerstone of this new paradigm.
ELIZA exemplifies the bazaar philosophy: an open framework where developers can build, experiment, and launch AI agents without constraint while contributing directly to the main protocol. Shaw’s open leadership style mirrors the ethos of his creation — AI Marc Andreessen, the AI partner powering the ai16z investment DAO.
We’re starting to realise that ELIZA is not a protocol. ELIZA is a movement.
Let’s explore the principles driving ELIZA’s design, the community it’s cultivating, and where value might ultimately accrue in this hyper-growth ecosystem.
We know which one we prefer.
At its core, ELIZA offers a modular architecture for creating AI agents that can seamlessly interact with both users and blockchain systems. While it shares its name with the iconic 1960s chatbot, this version of ELIZA is a bold reimagining. And much better looking, to be honest.
The soul of every ELIZA agent begins with its character file, a blueprint that defines the agent’s personality in rich detail. Think of it as crafting a digital persona, where agent developers can shape their agent’s identity through six key elements:
In ELIZA, character files are the equivalent of UI design for traditional software. It defines how users experience and engage with an agent.
By integrating built-in Retrieval-Augmented Generation (RAG) capabilities, ELIZA allows agents to access a knowledge base alongside its queries. This eliminates the complexity of maintaining personality consistency across platforms.
It frees developers to focus on what truly matters: crafting vibrant, memorable characters rather than getting bogged down by backend intricacies.
If the character file defines an agent’s essence, the Agent Runtime is its beating heart.
ELIZA provides an out-of-the-box framework for orchestrating everything from message processing to memory management and state tracking. This architecture allows developers to skip the drudgery of building infrastructure and instead focus on what makes their agent special.
Rapid prototyping and deployment have become much easier, enabling builders to iterate faster when building new AI experiences.
ELIZA’s action system is a fundamental shift from traditional AI frameworks. Here, every agent action—even sending a message—is treated as a discrete event. This approach splits decision-making into two stages:
This separation unlocks powerful capabilities, such as multi-stage workflows and robust validation processes.
For instance, an agent might identify a user’s intent to trade crypto, but the actual execution of the trade will only occur after passing stringent risk checks and verification steps. This makes it ideal for blockchain applications where security is crucial.
ELIZA’s providers enrich conversations by injecting real-time context, enabling dynamic and responsive agent behaviour.
Imagine a “boredom provider” that tracks user engagement during a conversation. If a user becomes repetitive or unengaging, the agent can mirror this by showing reduced enthusiasm—subtly mimicking real human interactions and making conversations feel more authentic.
This creativity extends further when providers work with evaluators, ELIZA’s reflection system. Evaluators analyze and extract key details from interactions, feeding them into a multi-layered memory architecture:
Providers then retrieve and reintroduce relevant details, making interactions with the agent contextually rich.
For example, if a user reminisces about selling their red Lamborghini a year ago, an ELIZA agent can later reference this while discussing their shiny new yellow Tesla. This interplay between memory and context elevates user interactions, creating agents that feel less robotic and more like actual companions.
Three core innovations in ELIZA highlight its forward-thinking approach to AI agents. Each demonstrates how its team envisions the evolution of autonomous agents in Web3.
Autonomous trading is a high-stakes game—one mistake can lead to catastrophic losses. Yet, as AI agents take on a larger role in Web3, their ability to execute transactions independently is becoming a must-have.
This emerging frontier, AgentFi, mirrors the pivotal role yield farming played in DeFi’s rise. Shaw and ELIZA tackle the inherent risks head-on with a robust two-tiered system: a Trust Engine paired with secure trading execution.
Token metrics captured in the Trust engine
The Trust Engine serves as the first line of defence, employing advanced validation checks to analyze multiple risk dimensions in real-time. From detecting rug pulls to assessing liquidity thresholds and holder distributions, the engine ensures every trade is rigorously scrutinised for potential pitfalls.
For instance, trades are restricted to tokens with a minimum liquidity of $1,000 and a market cap of $100,000. Holder concentration is monitored closely, rejecting tokens where any single entity controls more than 50%. These guardrails create a safety net, mitigating the risk of trading in compromised markets.
On top of this foundation, ELIZA’s position management system introduces dynamic risk controls that adjust trading sizes based on liquidity tiers. Low-risk trades are capped at 1% of the portfolio, while higher-risk opportunities may scale up to 10%. Total exposure is limited to 10% of the portfolio, with automatic stop-losses kicking in at a 15% drawdown. This structured framework balances seizing opportunities and maintaining tight risk management.
Trade execution itself is powered by Jupiter, a leading aggregator on Solana for optimal swap routing. Each transaction passes through multiple validation layers before execution.
An error recovery system kicks into action in the event of anomalies—like network disruptions, wallet imbalances, or unexpected market shifts. It halts active trades, closes risky positions, and alerts administrators, ensuring the system remains resilient under pressure.
"It's not just about giving an agent the ability to trade - it's about creating an entire system of checks and balances that prevents catastrophic failures."
— Shaw
What truly sets ELIZA apart for 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 trade performance, logging every recommendation and decision.
This data becomes the foundation for refining strategies over time, blending quantitative metrics with qualitative insights from the community’s recommendations (on Discord). The result is agents that don’t just execute trades but evolve, becoming smarter and more effective with each interaction.
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 leading distribution channel. However, integrating agents across multiple social platforms is not as easy. It demands significant development effort and ongoing maintenance. These challenges slow down deployment and scalability.
ELIZA solves this problem head-on with a comprehensive client package system to streamline multi-platform distribution.
ELIZA’s client architecture abstracts away the complexities of platform-specific implementations. With a standardised interface, developers can deploy their AI agents across Discord, X, Telegram, and custom REST API endpoints with minimal extra code. 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 like media processing, authentication, rate limiting, and error handling are managed internally by each client. For developers, this means less time wrestling with integrations and more time focusing on building innovative, high-performing AI agents.
By taking the headache out of multi-platform distribution, ELIZA empowers developers to scale their agents with ease, and they reach users wherever they are.
It’s distribution, simplified.
ELIZA’s plugin system makes it easy for developers to extend core functionalities and add custom capabilities to their agents.
While many developers create plugins tailored to their own use cases, the real power of this system lies in community sharing. By releasing plugins to the broader ecosystem, developers contribute to a growing library of out-of-the-box features, dramatically expanding what every AI agent on ELIZA can do.
The success of this approach is evident in the vibrant, “bazaar-style” development it fosters. Just to highlight some examples of community-driven plugins:
• Bootstrap Plugin: Essential conversation management tools.
• Image Generation Plugin: AI-powered image creation capabilities.
• Solana Plugin: Blockchain integration with built-in trust scoring.
• TEE Plugin: Secure execution environments for sensitive operations.
• Coinbase Commerce Plugin: Cryptocurrency payment processing functionality.
ELIZA’s plugin system is its strategic and platform advantage. By prioritizing extensibility, ELIZA has built a foundation for continuous growth and innovation, where:
Here’s the simple flywheel:
More devs building on ELIZA → more functionality (e.g. plugins) the framework enables → Even more devs building on ELIZA
The landscape for AI agents is evolving at hyper-speed. This means the ability to quickly integrate new capabilities will make or break a platform. ELIZA’s plugin system gives it the best chance of staying ahead of the curve, creating a self-reinforcing ecosystem where developers, users, and agents can all thrive.
Shaw and his team have been incubating some intriguing ELIZA agents, each showcasing the potential of AI in decentralised systems.
While these agents are still “young” in AI terms, with features and capabilities under active development, it hints at exciting possibilities.
Marc AIndreessen, one of the AI partners of ai16z, is a fascinating but enigmatic figure in the ELIZA ecosystem. His X account remains largely inactive, with only one post outlining ai16z’s thesis. However, according to Shaw, Marc is actively trading and yield farming—likely leveraging ELIZA’s Trust Engine and trading plugins.
Shaw has also hinted at Marc’s training process during a podcast interview, revealing that the AI is part of an alpha chat group populated by some of the best traders in the space. This suggests that Marc isn’t just a standard trading bot but an evolving agent learning from…human expertise.
In contrast to Marc’s quiet demeanour, Degen Spartan AI is the loud, brash sibling—an agent seemingly trained on the chaotic energy of 4chan, meme culture, and Crypto Twitter. His posts on X are a mix of random trading insights and irreverent commentary, reflecting a unique personality within the ELIZA ecosystem.
What separates it from Marc AIndreesseen is that it does have its own pump.fun token, currently at a $60M market cap. While he hasn’t yet begun trading yet, the foundation is clearly being laid for more ambitious interactions. His unpredictable nature makes him both entertaining and one to watch as ELIZA agents evolve.
While not a singular agent, The Swarm represents Shaw’s grand vision: a decentralized network of AI agents working collaboratively with humans and each other.
In this model, agents lead other agents, coordinate tasks, and interact transparently on social media. This deliberate transparency is designed to avoid hidden protocols and ensure public accountability.
Shaw believes that agent swarms are inevitable—and transformative.
We’re on the same page: agent swarms are set to drive the next wave of innovation, products, and attention in Web3 AI agents come 2025. Next year, we anticipate ELIZA agents stepping up, engaging in large-scale, cooperative activities that will redefine the role of AI in Web3.
The eliza framework from @ai16zdao has an opportunity to become the React-equivalent for ai agents.
An open source library that provides a basis for building ai applications, while also allowing anyone to build their own libraries on top of it.
Excited to keep working on it.
— cygaar (@0xCygaar)
8:02 PM • Dec 8, 2024
When evaluating ELIZA’s growth, the key metric is developer adoption. As a framework, ELIZA’s success hinges on the enthusiasm and contributions of its developer community.
On this front, ELIZA isn't just growing; it's exploding.
On its GitHub, the number of forks and stars (a proxy for developer interest) is approaching vertical acceleration, resembling the classic hockey stick trajectory.
Even more impressive is the surge in plugins and commits, reflecting a thriving, active ecosystem of contributors. As of December 12, it has 3,861 GitHub stars and 1,103 forks with 138 contributors. And 13,000+ members in the Discord.
For comparison with top existing open-source agent frameworks:
Adding further fuel to this growth, ai16z has launched a Creator Fund aimed at bootstrapping and rewarding developers building on ELIZA. This initiative was made possible through the generosity of Elijah, a significant ai16z token holder, who pledged to reduce his ownership from 16% to 5%, donating the difference to establish the fund. The Creator Fund could accelerate innovation, attracting new talent to the ecosystem.
However, while ELIZA’s framework holds immense value, where this value ultimately accrues is far from straightforward. Quite literally, it’s the billion-dollar question.
Currently, there’s an official $ELIZA token endorsed by Shaw, representing a personalisation of the ELIZA framework. You can even interact with ELIZA directly on their website. The token sits at a modest market cap of $66M.
However, the clearest winner so far of ELIZA’s growth is $ai16z, the investment DAO token, which has reached a staggering market cap of over $800 million. The community and investors appear to view $ai16z as the symbolic and practical representation of Shaw, ELIZA, and the broader vision they embody.
The story of ai16z begins as a fundraising mechanism for AI Marc Andreessen’s trading activities. Launched on DAOS.FUN in October 2024, the token raised 420.69 SOL during its initial offering. Under this model, the funds raised can be actively traded to grow the asset base, with profits accruing to token holders.
No single individual—not even Shaw—can mint additional tokens without a DAO vote. Token holders have governance rights, which allow them to propose and vote on initiatives and shape the direction of the DAO.
The fund has a built-in expiration date: October 25, 2025. All principal investments and profits will be distributed to ai16z token holders on this date. Whether this timeline remains fixed or gets extended will depend on how the ecosystem evolves over the next year.
Currently, ai16z has a net asset value (NAV) of $17.7M, primarily composed of its holdings in ELIZA tokens, degenai, and fxn. This means that the ai16z token ($0.80 at the time of writing) trades at a 50X premium to its NAV, which almost seems ridiculous at first glance.
However, an efficient market is rarely wrong. It reflects several other factors driving demand for the token.
AI agent platforms are a brand-new category, having emerged just a few months ago. The market is still grappling with fundamental questions: What is the true size of the AI agent opportunity? Where will value capture occur?
In this early stage of development, without standardised business metrics that can be compared, relative valuations often serve as benchmarks.
Currently, Virtuals Protocol, the leading Web3 AI agent launchpad, has a token valuation of $1.8B, making it the clear market leader. In comparison, ai16z sits in second place. Many believe ai16z has the potential to catch up with—or even surpass Virtuals—if ELIZA continues to power the creation of more useful and innovative AI agents. Even by just purely riding on mindshare and investor/retail interest.
But it's not a given; the competition is heating up. In our opinion, it’s a race to the top, the best kind of race.
The AI Agent Infrastructure Wars are heating up.
@truth_terminal@AndyAyrey announces Loria: infinite backrooms on steroids
Multiple agents with the ability to interact with each other & stream data to themselves for fine-tuning.
@0xzerebro launching their ELIZA competitor… x.com/i/web/status/1…
— 7213 | Ejaaz (@cryptopunk7213)
2:24 PM • Dec 11, 2024
Monetizing open-source frameworks is notoriously difficult.
For ai16z, the key driver of value in the future will likely come from agent tokenomics: AI agents launched on ELIZA contributing a percentage of its tokens back to the ai16z DAO. In this sense, the price of the ai16z token should reflect a % of the aggregate future value generated by all agents built on the ELIZA framework.
Is the future value $10M? $100M? $10B? It’s anyone’s guess at this point, with so many unknowns, but the growth trajectory of ELIZA makes us lean on the optimistic side.
Today, contributions to the ai16z DAO are voluntary, with several projects donating between 1% and 10% of their tokens. Additionally, users deploying their AI agents on Vvaifu (a popular community launchpad for ELIZA agents) currently pay a fee of 1.5 SOL plus 5% of the agent’s token supply if they use the ELIZA framework. These contributions can be tracked on the ELIZA Observatory.
There’s speculation that ai16z could soon debut an official ELIZA agent launchpad, enforcing token contributions at the smart contract level for agents using the platform. However, as an open-source framework, ELIZA can still be used independently, meaning not all projects will necessarily tie themselves to ai16z.
The original vision of ai16z was to be a smart autonomous trader, with Marc AIndreessen (AI) at the helm. As far as we know, Marc only started trading recently, and there are scant details, so it’s hard to evaluate how good of a trader the AI will be.
What’s interesting, however, is the approach it is taking.
ai16z is building a “marketplace of trust”. In this virtual ecosystem, AI agents take insights from the community, simulate trades, and adjust their trust scores in real time based on how well their recommendations perform. The whitepaper for the marketplace is expected to come out before the end of the month.
The goal is to create AI agents capable of operating autonomously and securely within a self-reinforcing system of transparency and accountability. The Trust marketplace serves as a testing ground. While no actual trades occur initially, this environment allows agents to refine their capabilities safely, with a view towards live trading at some point.
Trust scores range from 0 to 1 (normalized to 100) and are a public badge of reliability, showcased on a leaderboard for all to see. User recommendations feed into the system, with more credible users (those with higher trust scores) carrying more weight.
It’s a feedback loop built on logic: agents simulate trades, users are evaluated on the outcomes, and everyone’s trust scores are updated accordingly. Over time, the system gets smarter, more reliable, and, well, more trustworthy.
Adding a social layer is the public trust profile, where agents and users are incentivized to build their reputations. Community moderation ensures accountability and transparency
In crypto, speculation often frontruns product-market fit, revenue generation, and long-term value capture. For ai16z, much of its current valuation can be attributed to the mindshare it has captured in the burgeoning AI agent ecosystem.
ai16z has positioned itself as a top-of-mind AI agent framework with a thriving developer community and a fast-growing ecosystem.
Here’s the ai16z narrative: a “cracked” dev team is actively shipping tutorials, creating innovative agents, and spearheading development in the space.
The team’s bi-weekly AI agent dev school sessions on X have further cemented its reputation. The first session alone drew over 12,000 live viewers, showcasing the massive interest in building AI agents on ELIZA.
Right now, ELIZA is firmly planted in the Solana ecosystem, but its rapidly growing plugin system is laying the groundwork for a multi-chain future.
ELIZA’s true potential lies in Shaw’s vision of “the swarm”: a decentralized network of AI agents pooling resources and collaborating across ecosystems. This swarm effect could establish a durable competitive advantage akin to the liquidity depth that drives value in DeFi protocols.
Ultimately, the goal is to create open standards for agent communication, mirroring the transformative impact of ERC-20 in token interoperability.
For all its potential, ELIZA faces one of the toughest challenges in tech: making open-source frameworks sustainable. If the community loses interest (e.g. if the token price goes on a sustained downward spiral or a new shiny thing arises), development halts/slows, and it can become hard to catch up.
There’s also a lot of messiness when the community is directly involved in the codebase and pushing changes rapidly — instability, poor documentation, frequent crashes and bugs that undermine the user experience.
The framework’s greatest opportunity lies in crypto-native incentives.
If ai16z can design effective tokenomics that reward ELIZA contributors and align them towards its success, it could bring traditional open-source projects onto crypto rails. Imagine GitHub meets DeFi, where contributors are rewarded not just with clout but with real, tangible financial value.
In our view, ELIZA is not just another AI agent framework competing with LangChain or CrewAI—it’s something far more ambitious.
It’s a living embodiment of the bazaar philosophy, where open-source development thrives in an ecosystem powered by collaboration and creativity.
With its modular architecture, innovative trust engine, and expansive plugin system, ELIZA is an experiment on how AI can reshape open-source development itself.
What makes ELIZA truly exciting is its position 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 won’t just shape the future of AI agents—it could fundamentally redefine how we build software.
For now, the bazaar is bustling.
Cheers,
ChappieOnChain & Teng Yan
You can follow ChappieOnChain on X for insightful and sometimes spicy takes on AI agents.
The authors of this report may personally hold a material position in the tokens mentioned within. This report is intended solely for educational purposes and does not constitute financial advice. It is not an endorsement to buy or sell assets or make financial decisions. Always conduct your own research and exercise caution when making investment choices.
