Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
This page explores how decentralized AI and blockchain technology empower users to take control of their digital experiences.
We're setting the stage by outlining how Web3 redefines user roles and data ownership compared to the traditional internet.
As the digital world evolves from Web2 to Web3, everyday users face a transformation in how their data is utilized and their voices are heard. In the traditional model, centralized platforms have dominated, often exploiting user data without equitable returns. Now, a new paradigm is emerging—one where users are no longer passive data providers but active participants in a decentralized ecosystem.
Decentralized AI on Web3 offers a radical shift. By leveraging blockchain technology and democratized machine learning, platforms can offer personalized experiences without compromising user control. This approach promises transparency, accountability, and, most importantly, the restoration of user agency in an environment where AI works for you, not against you.
Below are the current problems that users encounter in the digital landscape, highlighting the consequences of centralized control and opaque AI systems.
In the current digital landscape, your data fuels vast profit engines. Major platforms harvest and monetize user data, often without your explicit consent or a fair share in the resulting economic value. This model reduces your personal information to mere statistics in a game designed to maximize profit margins.
The inner workings of machine learning algorithms remain largely opaque. Without insight into how decisions are made, users are left in the dark about why certain content is served to them, leaving them vulnerable to manipulation and bias. This lack of transparency undermines trust and erodes user control over personal digital experiences.
While personalized content can enhance convenience, it often comes at the cost of diversity. Algorithms that tailor information based solely on past behavior risk limiting your exposure to new ideas and perspectives. This narrow focus can reduce the richness of your online experience and stifle intellectual growth.
The over-personalization of content contributes to the formation of digital echo chambers. As platforms continuously refine what they show based on your previous interactions, you might find yourself confined within a bubble of similar viewpoints. This fragmentation not only hinders healthy public discourse but also deepens social divides.
And finally, here are solutions Deep3 offers to overcome current digital challenges by restoring user control and ensuring fair compensation.
Deep3 is built on the principle that users deserve a direct say in how AI technologies are developed and deployed. Our decentralized governance model empowers you to participate in critical decisions, ensuring that the tools you use reflect your needs and values rather than those of a centralized authority.
We believe that transparency is key to rebuilding trust in digital platforms. With Deep3, the algorithms driving your experience are open for inspection and understanding, enabling you to see exactly how your data influences the content and services you receive.
Deep3’s approach to personalization strikes a balance between relevance and respect for your data. By using ethical machine learning practices, we tailor experiences to your preferences without commodifying your personal information. This creates a digital environment that’s both engaging and respectful.
Our platform is designed for everyone. Whether you’re a tech-savvy user or someone just exploring the digital space, Deep3’s intuitive interface ensures that advanced AI tools are accessible to all. We strive to eliminate the technical barriers that have traditionally kept many users on the sidelines of innovation.
Imagine a digital ecosystem where your contributions are recognized as valuable assets. Deep3 enables you to benefit directly from your data, transforming passive information into tangible economic rewards. This model redefines the relationship between users and platforms, placing fair compensation at its core.
At Deep3, we’re committed to creating a balanced environment where both innovation and user rights flourish. By fostering an ecosystem that rewards your participation, we promote a more equitable digital future. Here, the value generated by AI isn’t hoarded by a few tech giants—it’s shared with everyone who contributes to its success.
This page explores how decentralized AI transforms business operations in Web3.0 by offering innovative, ethical, and efficient solutions for data-driven growth.
We see an evolving landscape of Web3, where decentralized technology redefines how businesses operate and compete.
The shift from traditional, centralized models to decentralized ecosystems is reshaping every facet of business—from data management to customer engagement. As businesses move away from siloed systems, they are embracing open, collaborative models that enable more agile and transparent operations.
Decentralized AI is unlocking new potential by combining advanced machine learning with blockchain technology, driving efficiency and innovation. By integrating these technologies, companies can harness the benefits of data-driven insights while ensuring greater accountability and security.
Below are the key obstacles that businesses encounter under conventional models and the limitations of current AI/ML practices.
Traditional data management systems often restrict access and fragment valuable information, limiting the ability to extract meaningful insights. Companies struggle with disjointed data silos, which hinder efforts to monetize data effectively and reduce the overall strategic value of their digital assets.
Opaque AI algorithms and evolving regulations create uncertainty, making it difficult for businesses to innovate while staying compliant. The lack of clear, transparent processes erodes trust among customers and regulators alike, posing significant barriers to adopting new technologies.
Continue readint to see how Deep3’s platform addresses these challenges by delivering turnkey AI/ML solutions designed specifically for the Web3 ecosystem.
Deep3 offers plug-and-play solutions that allow businesses to incorporate AI into decentralized applications without the typical integration headaches. Our platform simplifies the implementation process, ensuring that innovative AI tools can be adopted quickly and efficiently across your dapps.
By fostering an ecosystem that values fair compensation, Deep3 turns user data into a strategic asset without compromising ethical standards. Our approach enables businesses to generate sustainable revenue streams while maintaining transparency and trust with their user base.
Deep3’s technology not only personalizes user experiences but also drives deeper engagement, helping businesses stand out in a crowded market. By leveraging data-driven insights, companies can build stronger relationships with their customers and gain a significant competitive advantage.
An intrepid team of experienced developers, researchers and business leaders.
We're a diverse group of tech-obsessed people that has been working at the intersection of blockchain and AI since long before your grandma had a ChatGPT account. Founded by industry veterans from Google and Apple, we believe that these two technologies can help each other tremendously, which in turn, will help the rest of us.
Deep3 Labs is guided by a few simple, but important core beliefs. Not long ago in Web3, many of these beliefs were often seen as futuristic or even eccentric. But over time we've seen them become increasingly normalized, which let's us know that we're on the right track.
In order to attract the "next billion users", blockchain developers must learn how to create the same ultra-rich, highly-personalized user experiences that today's internet users have come to expect. Facebook would be a chat forum from the mid-90s and Google would be the "online yellow pages" without their use of machine learning. Algorithms, specifically machine learning and artificial intelligence are single-handedly responsible for creating the kinds of user experiences that have brought billions of people online...and kept us there.
In order to not "end human civilization as we know it", AI engineers must explore new governance and development systems that ensure AI is always working for us, and never the other way around. Right now, a very small number of people, generally executives and engineering leaders, enjoy a monopoly on all decision-making power over how AI/ML is used in the apps we love. What's more, there's no clear pathway for us to even see the decisions they make in the development and deployment process. Perhaps for the first-time ever, blockchain technology offers solutions to these problems that are both practical and scalable.
It's no secret that even the most advanced blockchains and the dapps that have been built on them aren't capable of attracting mainstream users. They can be slow, difficult to interact with, and worst of all, contemporary Web3 user experiences are monolithic — they behave exactly the same for every user that connects. Algorithms can be used to solve all of these problems, and research is already underway to prove just that. AI can be used to optimize how loads are balanced when transaction volume spikes. It's capable of abstracting convoluted aspects of constructing contract calls and transactions such that users can simply describe their intent in plain english. And, as evidenced by the work we're doing at Deep3 Labs, they can be used to create user experiences that are tailored to the unique preferences held by each of us when we connect our wallet to a dapp.
Our goal is simple, but ambitious:
To bring meaningful AI/ML capabilities to Web3 while sharing the control AI requires and the value AI generates with the everyday users most affected by its actions and most crucial to its creation.
We also took the time to explain this mission in video form:
And in words (but only as many as can fit on one page):
Now, unpacking this mission, and explaining exactly how we expect to do it, will be quite the journey. These docs will guide you along the way, but if you've come here with a more specific question, check out the links below.
Quickly get the gist of "what", "why" and "who deep3"?
Our vision for AI/ML on Web3 consists of three main ecosystem components.
Whether you're a developer or a user, we're building stuff to help you unlock the power of AI on Web3.
We're building easy-to-use Web3 AI/ML tools and dapps that empower everyone to share in the benefits and control of AI.
Our platform allows blockchain developers and blockchain users to harness the power of machine learning with little or no code. We train prediction models on the data contained in the blockchain's ledger, expose the output of those models in no-code dapps (for users) and a low-code API (for developers), then enable everyone to share in its benefits and control. That means you'll have a real say in how your favorite dapps use AI and a real share of the value AI creates thanks to your user data.
Soon, you'll even be able to build and deploy your own advanced machine learning models using a simple chat interface, enabling the breadth of possible applications of machine learning in Web3 to grow rapidly. Expert data scientists will also have a forum to train and share ML assets, allowing them to monetize this in-demand skill set.
Machine learning is already being used across the internet to create enormous value, that is, everywhere except in Web3. We see a future where thoughtfully designed AI/ML tools can be used in Web3 to:
Deep3 empowers everyone to have a direct say in how AI evolves on Web3. Our decentralized governance model ensures that control is distributed among users, developers, and data scientists rather than concentrated in a single entity. This transparent decision-making process allows the community to collectively shape the rules and policies guiding AI tools. By embedding governance into the platform, we create a system where every stakeholder can influence how algorithms operate and how their data is used, ensuring that the benefits of AI are aligned with the collective interest.
Deep3 revolutionizes how value is generated and distributed in the blockchain ecosystem. Our AI tools transform user data into tangible economic benefits, ensuring that the rewards from improved dapp performance and innovative service offerings are shared with the community—not just captured by centralized platforms. By integrating value sharing into our framework, Deep3 enables every participant to earn a direct stake in the success of blockchain applications. This model not only incentivizes continued engagement and loyalty but also drives a virtuous cycle of innovation, making blockchain dapps more efficient, user-friendly, and economically rewarding for everyone involved.
Part of our go-to market strategy centers on building flagship dapps in key sectors where AI integration in Web3 user experiences is still in its infancy. By pioneering AI-enhanced applications in these lucrative verticals, we secure a first-mover advantage that positions Deep3 as a trailblazer in transforming digital interactions. This approach not only demonstrates the tangible benefits of advanced machine learning in real-world use cases but also drives demand for our turnkey tools, inspiring other builders to explore and expand the possibilities of AI within Web3.
We've already built three AI-powered dapps to show what's possible when you build with our tools.
A Note on this Document's Construction
A machine learning model is a smart system that learns from data to predict outcomes or help you make decisions. Think of it like a recipe: you provide the ingredients (data), and the model mixes them using math to create something useful—whether that's predicting trends or identifying patterns.
In Web2, these models have been the secret sauce behind popular tools such as recommendation engines, search results, and targeted ads. They analyze vast amounts of data to deliver personalized experiences and efficient services. With Deep3, we adapt these proven techniques to blockchain data, opening up exciting new applications in trading, advertising, and security.
At Deep3, our models are not hidden behind the scenes—they are built into products that deliver real value. In trading, our models can forecast market trends and help optimize investment strategies. In advertising, they tailor content to the right audience at the right time, while in security, they work to detect potential threats before they become problems.
By packaging these advanced machine learning capabilities into easy-to-use products — whether in our dapps or for other builders to use via our API — , we empower both individuals and businesses to make smarter decisions on Web3. Whether you’re a trader looking to stay ahead of the curve, a marketer striving for better targeting, or a security team aiming to protect your assets, Deep3’s models transform raw data into actionable insights that drive success in the world of Web3.
StakeSage-L is our liquid staking lifetime value (LTV) estimation model built on the Ethereum network. This model accurately predicts the total lifetime staking amount that a blockchain address is likely to commit in the future, with a precision of up to 0.01 ETH. By providing actionable customer intelligence, StakeSage-L empowers staking platforms to offer revenue-maximizing incentives to high-value users and optimize their purchase funnels, all while enhancing the overall user experience.
Designed as a turnkey solution, StakeSage-L brings advanced AI capabilities to your application without the need for building an in-house data science team. Its precise predictions enable dynamic adjustments in staking offers, ensuring that both platform operators and their customers benefit from tailored incentives that drive long-term engagement and improved operational metrics.
StakeSage-L delivers high-resolution insights by accurately estimating the lifetime staking value for each address. It identifies high-value stakers with pinpoint accuracy, allowing platforms to design and deploy dynamic staking terms that reward users based on their predicted commitment levels.
In addition, the model is engineered to enhance customer intelligence by providing granular data on user behavior. This supports the optimization of marketing strategies and product design, leading to more personalized user experiences. The model’s ability to process large volumes of on-chain data ensures it remains robust, scalable, and continuously updated with the latest market dynamics.
StakeSage-L is powered exclusively by on-chain data from Ethereum. For training, the model leverages a comprehensive dataset that includes 212,293 wallets, 2,723,000 token features, and 344,729 transactions, with predictions scored for 194,318 wallets. These detailed inputs enable the model to capture the intricate nuances of staking behavior, from initial participation to lifetime commitment.
This high-fidelity data set ensures that the model’s predictions are grounded in real, verifiable blockchain activity. By using such specific and extensive inputs, StakeSage-L is able to deliver insights that are both accurate and actionable, providing a solid foundation for forward-looking staking strategies.
In developing StakeSage-L, we rigorously evaluated multiple regression techniques including Linear Regression, Decision Tree Regressors, Random Forest Regressors, XGBoost, and CatBoost. CatBoost emerged as the superior technique, outperforming XGBoost by 12% and linear regression by 61%, thanks to its effective regularization methods that better suit our training set's characteristics.
The model employs advanced feature engineering to quantify key staking behaviors and predict future staking amounts with a high degree of precision. By blending state-of-the-art algorithms with robust on-chain data, StakeSage-L transforms raw transaction data into a reliable prediction of lifetime staking value, providing deep insights into user behavior for dynamic decision-making.
StakeSage-L boasts a prediction accuracy within 0.01 ETH, making it one of the most precise models in its category. This level of accuracy is achieved through rigorous model development and extensive testing, ensuring that the predictions are both reliable and actionable for high-stakes financial decisions.
To validate the model, we employ out-of-sample testing, where the model is evaluated on data it has never seen before. This process simulates real-world conditions and confirms that the model maintains its accuracy beyond the training dataset. By comparing predicted values against actual lifetime staking amounts, we continually refine StakeSage-L to ensure it meets the high standards required for strategic decision-making.
StakeSage-L is primarily used to drive LTV-driven dynamic staking terms, enabling platforms to strategically offer incentives tailored to each user’s predicted lifetime value. This approach helps maximize average order value (AOV) and secures long-term loyalty from high-value stakers.
Beyond optimizing staking offers, the model enhances overall customer intelligence. By providing precise insights into user behavior, StakeSage-L supports targeted marketing campaigns, refined product design, and improved operational strategies. This comprehensive view of user engagement allows platforms to dynamically adjust their user interfaces and purchase journeys, ultimately leading to better customer retention and increased profitability.
HODL-C1 is our token investor scoring model designed to identify blockchain addresses with a high likelihood of becoming long-term token holders. Deployed on Ethereum, Polygon, and BSC, it delivers actionable customer intelligence by accurately predicting investor behavior. With nearly 90% prediction accuracy, HODL-C1 enables blockchain businesses to optimize operations—whether refining sales strategies or enhancing security—by understanding and engaging their customer base more effectively.
Developed as a turnkey solution, HODL-C1 brings advanced AI capabilities directly to your application without the need for an in-house data science team. Its insights empower businesses to offer tailored, dynamic benefits, such as optimized IDO terms, while providing an enhanced user experience that mirrors the personalized touch found in Web2.0 services.
HODL-C1 offers high-resolution targeting by delivering address-specific customer intelligence. Its predictions allow launchpads to design tiered benefits and dynamic offering terms that reward long-term holders fairly, regardless of their wallet size. This means more inclusive and effective customer segmentation, helping you cater to a diverse investor base.
Other key features include its ability to process massive volumes of on-chain data, enabling the model to stay current with evolving blockchain behaviors. With an intuitive design, HODL-C1 transforms complex data into clear, actionable insights that drive smarter business decisions.
HODL-C1 is powered exclusively by on-chain data, drawing from Ethereum, Polygon, and BSC networks. It leverages extensive datasets—including millions of wallet records, billions of transactions, and comprehensive token features—to develop a robust understanding of investor behaviors. For example, on Ethereum alone, the model was trained using over 49 million wallets, nearly 900 million token features, and more than 1.6 billion transactions.
This rich dataset enables HODL-C1 to capture the nuances of wallet activity, from trading frequency to overall token holding patterns. The breadth and depth of the data ensure that the model’s insights are both precise and scalable, providing a reliable foundation for its high-accuracy predictions.
HODL-C1 is powered exclusively by on-chain data, drawing from Ethereum, Polygon, and BSC networks. It leverages extensive datasets—including millions of wallet records, billions of transactions, and comprehensive token features—to develop a robust understanding of investor behaviors. For example, on Ethereum alone, the model was trained using over 49 million wallets, nearly 900 million token features, and more than 1.6 billion transactions.
This rich dataset enables HODL-C1 to capture the nuances of wallet activity, from trading frequency to overall token holding patterns. The breadth and depth of the data ensure that the model’s insights are both precise and scalable, providing a reliable foundation for its high-accuracy predictions.
HODL-C1 stands out with nearly 90% prediction accuracy across Ethereum, Polygon, and BSC. This impressive performance metric reflects the model’s ability to correctly identify addresses that are poised to be long-term token holders, a critical capability for optimizing dynamic initial DEX offering (IDO) terms.
To ensure that HODL-C1’s impressive performance is robust and not simply a result of overfitting to its training data, we conduct extensive out-of-sample testing. This means we evaluate the model on fresh, unseen data—separate from the data used during training—to simulate real-world conditions. By confirming that the model maintains its high prediction accuracy on this new data, we build confidence in its ability to perform reliably in live environments.
HODL-C1 is primarily used to generate personalized, data-driven recommendations for blockchain businesses. One key use case is optimizing IDO terms by identifying high-value investors, enabling launchpads to offer the most compelling benefits to wallets with the greatest potential for long-term engagement. This targeted approach helps both the launchpad and its fundraising teams secure a more committed investor base.
Beyond optimizing token offerings, HODL-C1 can serve as a powerful tool for customer intelligence. By providing detailed insights into investor behavior and segmentation, the model supports more informed decisions in marketing, product design, and operational strategies. Whether you’re looking to refine your sales approach or enhance overall security, HODL-C1 delivers the precision and depth needed to understand and serve your customer base effectively.
We learned a lot from how machine learning and artificial intelligence worked, and didn't work on Web2. It created some of the world's most powerful companies, but we'd also argue it created some of our thorniest problems.
The mission of Deep3 Labs is to reimagine the way Web3 businesses will create and deploy vital AI/ML technology for the enhancement of blockchain dapps, products, and user experiences in a manner that:
Restores agency to online users (through decentralized governance)
Lowers the barriers to entry for builders (through turnkey implementation)
Recognizes the monetary value of online user data (through sustainable value sharing)
Our collective motivation has both professional and personal origins because each of us are both creators and consumers of AI/ML technologies. The importance of this fact can’t be understated in terms of morale, engagement, and recruitment.
We view the growing prevalence of data- and technology-related legislation as a figurative “shot across the bow”. While well-intentioned, laws such as GDPR and CCPA – both of which primarily restore consumers’ data rights – would be extremely problematic if similar frameworks were applied to machine learning production processes, which we're already starting to see. On the one hand, providing an individual the means to “opt-out” of a machine learning process is technically far more complex than being excluded from data harvesting or warehousing. On the other – and more importantly–, these laws tend toward prohibition rather than a mutually-beneficial collaboration between users and platform creators, which given the nascency of the AI/ML space, could dramatically hinder future discovery. AI/ML technology will be central to solving some of the world’s most important problems, but only if we are free to continue exploring.
The above has served as a powerful “rallying cry” in Deep3 Labs’ early recruiting efforts.
Our personal motivations may in fact be far more compelling as they generalize to any internet user, regardless of their AI/ML knowledge base. We live in a world where the largest corporations on Earth consume user data as their primary input to production and, for a variety of reasons, no mechanisms exist to ensure a fair price is being paid for this resource. Furthermore, the value-generating potential of user data only stands to accelerate in the future as artificial intelligence gains increasing commercial traction. And finally, growing disparities in wealth and influence have become undeniable in nearly every country and culture the world over, a phenomenon exacerbated by the growing use of AI/ML technology. Considered together, it‘s reasonable to expect that existing negative externalities will only intensify in the future, such as the specific social impacts of screen-time maximizing machine learning objectives or racial biases observed in model outcomes, as well as a more general perpetuation of inequality in the markets they operate in.
These problems touch all of our personal lives in profound ways, thus cementing this as a powerful driving force behind our work.
If you're ever the unfortunate victim of being cornered by one of us at a cocktail party or a conference, you're sure to hear some ranting along the following lines.
If big tech ran on an ecosystem with the features and objectives of the Deep3 platform, Ms. Haugen wouldn't be famous, and we suspect that'd be fine by her.
In short, Ms. Haugen revealed to the world, and the United States Congress, that the executives at Facebook and Instagram knew that their algorithms were destroying young lives. But, they chose to continue operations as-is, and in many ways, double-down on some of the most detrimental elements of their design.
The relentless drive for ad dollars has warped the way our digital platforms operate. Machine learning algorithms are fine-tuned not to enrich our online experiences, but to maximize clicks and engagement—even if that means promoting content that’s sensational or divisive. In this model, every interaction is distilled into data points, reducing your online life to metrics that serve profit margins, not your well-being.
Equally troubling is how personalization traps you in a bubble of your own making. By curating content that mirrors your past behavior, these algorithms limit your exposure to new ideas, deepening divisions and reinforcing biases. Over time, this self-reinforcing cycle turns our digital spaces into echo chambers, fragmenting public discourse and undermining the diversity of thought essential for a healthy society.
CLUSTR-1 is a cutting-edge clustering algorithm developed primarily for trading applications and forms the core of our trading dapp, Hoku. Trained initially on Base and Ethereum on-chain data, CLUSTR-1 is designed to +, enabling personalized trading recommendations. Our roadmap includes expanding its capabilities to support additional networks such as Solana, BSC, and Arbitrum, opening up a broader range of use cases.
The algorithm leverages the power of unsupervised learning to identify patterns in trading behaviors without relying on predefined labels. By characterizing wallets based on risk tolerance and transaction dynamics, CLUSTR-1 delivers insights that empower traders with tailored, actionable recommendations.
CLUSTR-1 harnesses advanced clustering techniques to uncover hidden patterns in trading data, ensuring that users receive personalized and dynamic insights. One of its standout features is the development of embeddings that capture intrinsic wallet characteristics—such as age and transaction frequency—alongside comprehensive trading histories.
In addition to its robust clustering capabilities, CLUSTR-1 employs dimensionality reduction to create a three-dimensional manifold representation of the data. This 3D representation feeds directly into Hoku’s unique explorer interface, making it easy for users to visualize and interact with complex trading data in an intuitive way.
The training of CLUSTR-1 is built entirely on on-chain data, ensuring that all insights are grounded in real, immutable blockchain records. The algorithm processes detailed wallet metrics, including wallet age, transaction frequency, and complete trading histories, to derive a comprehensive view of each wallet's behavior.
By focusing on this high-integrity data, CLUSTR-1 is able to generate reliable embeddings that accurately reflect both the innate attributes of each wallet and its broader trading activity. This approach guarantees that our model remains both transparent and robust in its analysis.
While we maintain confidentiality around the specific techniques employed, the core of CLUSTR-1 revolves around generating embeddings that succinctly capture a wallet’s essential traits and trading history. These embeddings serve as the foundation for advanced clustering methods, which group wallets based on their risk tolerance and behavior patterns.
The final step in our process involves applying dimensionality reduction to transform the high-dimensional data into a three-dimensional manifold. This manifold not only simplifies the complexity of the underlying data but also powers the innovative explorer interface in Hoku, providing users with an engaging and intuitive visualization of the trading landscape.
As an unsupervised learning model, CLUSTR-1 does not conform to traditional accuracy metrics. However, we measure its effectiveness using silhouette scores, which consistently hover around 0.7, indicating strong and meaningful cluster separation.
Beyond these quantitative metrics, our backtesting has shown that many of the trading recommendations generated by CLUSTR-1 have achieved returns exceeding 100x. This impressive performance underscores the model’s practical value and its potential to revolutionize trading strategies.
The primary use case for CLUSTR-1 today is driving personalized trading recommendations within Hoku, allowing users to optimize their strategies based on nuanced insights into wallet behavior. By tailoring advice to individual risk profiles and trading histories, the model helps traders navigate the market with greater confidence and precision.
Looking ahead, the versatile nature of CLUSTR-1 positions it for expansion into other domains. Future applications could include personalized experiences in gaming, SocialFi platforms, and even the emerging market for collectible NFTs, demonstrating the wide-ranging potential of our clustering approach in the broader Web3 ecosystem.
The people behind the work happening every day at Deep3 Labs.
Data scientist and economist with over 15 years experience. Former Google Senior Analyst, biotech CTO, and Fortune 100 Director of Econometrics. 9 years of experience in blockchain and a graduate of the Binance Labs Incubator Program.
Highly experienced full stack engineer and team lead with 25 years experience across supply chain, finance and sales. Developed multi-million dollar custom IT systems for Fidelity Investments and The Boston Beer Company.
Data scientist with a specialization in cryptocurrency analysis, particularly in trading and investor behavior analysis. Designed and developed many profitable arbitrage and prediction models in the Ethereum ecosystem 3 years of experience in blockchain.
Creative full stack developer with 7+ years of experience, blending a biomedical engineering background with expertise in software development. Proven skills in e-commerce and SaaS development, with 3.5 years focused on blockchain frontends.
Data analyst and researcher with over 6 years of experience in data analysis, operational analytics and field research. Experience in both the banking and energy sectors as well as Masters of Science in Economics. 2 years of experience in blockchain.
Product and User Interface designer with experience across multiple verticals and a focus on functional applications and design. Former 5ireChain and Deeplink creative designer. 3 years of experience in blockchain.
Global supply chain and large customer relationship management expert specializing in technology sectors. Former Apple Global Supply Manager for Special Products. 6 years of experience in blockchain.
Community marketing expert and former publication manager with over 7 years of experience in service based industries working with both in-person and remote groups. Built and managed successful crypto mining operations. 6 years of experience in blockchain.
USAF Veteran, 6 years in Aircraft Electronics, 12+ years in IT systems and InfoSec, handling Secret Classified cryptographic codes pertinent to US National Security. Former Bitcoin & Ethereum Miner circa 2017, Blockchain Technology analyst & consultant
The people behind the people at Deep3 Labs.
Highly credentialed researcher with a PhD and MSEE in Electrical Engineering and Computer Science from MIT. Has produced 2 patents, 6 textbooks, 9 book chapters, 3 encyclopedia entries and 100+ technical papers.
Distributed applications architect with over 30 years experience. Solidified.io smart contract auditing co-founder, lead smart contracts trainer at B9Labs, and Top 50 most influential Canadians in Crypto. 7 years experience in blockchain.
Finance executive and private investor with over 25 years of experience. Strategic advisor to Trammell Venture Partners, Episode Six and Vertalo. B.S. Economics from Wharton and MBA from Stanford. 6 years of experience in blockchain.
Corporate attorney and management consultant with over 25 years of experience, specialty in startups. Managing partner at Root Protocols, a Web3 incubator. 6 years of experience in Blockchain.
International economist, business leader and diplomat with over 20 years of experience. Specializes in mechanism design and game theory. Honorary Hungarian Consul for the State of Connecticut and Director of the Central European Institute.
Data-driven marketing leader with 15 years experience. Masters in Marketing Analytics from NYU, with specialized expertise in early-stage growth strategies. Demonstrated track record of success growing revenues to $50M+ in two years.
StakeSage-C is our liquid staking conversion rate estimation model designed for the Ethereum network. It accurately predicts the likelihood that an Ethereum wallet—without any previous liquid staking deposit (LSD) interaction—will perform its first liquid staking transaction. With over 95% accuracy on out-of-sample test data and availability for more than 130 million addresses, StakeSage-C empowers platforms to enhance customer targeting and optimize user conversion strategies.
Developed as a turnkey solution, StakeSage-C enables you to maximize return on ad spend (ROAS), dynamically adjust purchase funnels, and gain deep customer intelligence without the complexity and cost of developing an in-house solution.
StakeSage-C delivers actionable insights with high precision by identifying wallets that are most likely to convert to liquid staking. Its capabilities help improve marketing performance by focusing on users with a high probability of conversion, thereby ensuring advertising budgets are spent effectively. The model also supports dynamic purchase funnels by tailoring the user journey based on conversion readiness, guiding users directly to purchase or to further information as needed. Moreover, StakeSage-C provides enhanced customer intelligence through rich, machine learning–powered insights that enable better segmentation and personalized messaging.
StakeSage-C is powered exclusively by on-chain data from Ethereum. The model was trained on an extensive dataset that includes 483,050 wallets, 6,279,650 token features, and 16,374,701 transactions. This robust dataset has been applied to score over 130,960,187 wallets, ensuring that the model captures the full complexity of wallet behaviors. Such detailed and specific inputs enable StakeSage-C to provide precise predictions on conversion likelihood.
For the development of StakeSage-C, we rigorously evaluated several binary classification frameworks, including logistic regression, XGBoost, and artificial neural networks (ANNs). Logistic regression significantly underperformed compared to the other methods, while ANNs did not provide enough gain to justify their additional deployment and computational costs. Ultimately, XGBoost emerged as the optimal choice, balancing high accuracy with efficiency.
The model uses a multivariate prediction framework that analyzes complex wallet characteristics derived from on-chain data. This sophisticated yet efficient approach allows StakeSage-C to capture intricate user behaviors and accurately predict conversion likelihood, ensuring that its predictions are both robust and actionable.
StakeSage-C has demonstrated over 95% accuracy on out-of-sample test data, confirming its ability to generalize well to new, unseen wallet addresses. Out-of-sample testing means the model is evaluated on data it did not see during training, which is a strong indicator of reliability in real-world conditions.
Beyond these controlled tests, a two-month live validation study concluded in January 2024 found that StakeSage-C correctly identified nearly 70% of new stakers 30 days in advance—representing a 13% improvement over the baseline. These real-world results underscore the model’s practical effectiveness, highlighting its potential to significantly boost marketing and conversion outcomes in live environments.
StakeSage-C is designed to maximize marketing ROI by precisely targeting high-value users, ensuring that advertising efforts are focused on those most likely to convert. By accurately predicting conversion behavior, the model supports dynamic purchase funnel optimization—directing users to purchase when they are ready or guiding them to additional information when needed. In addition, the deep customer intelligence provided by StakeSage-C enables more effective segmentation and personalization strategies, which can significantly enhance overall user engagement and drive long-term business growth.
DeepShield-HFT is our specialized security model designed to identify high-frequency trading (HFT) addresses on Ethereum. With an accuracy of over 95%, it helps dapp developers and users alike distinguish between organic trading activity and machine-controlled bots that can disrupt market integrity. Although DeepShield-HFT currently focuses on historical detection, its architecture can be extended to support real-time use cases in the future.
Built on the same foundational architecture as our HODL-C1 model, DeepShield-HFT benefits from the rigorous development and testing methods that have proven successful in our other solutions. This ensures a robust and scalable approach to detecting rapid trading patterns that might otherwise go unnoticed.
DeepShield-HFT excels at flagging addresses that consistently perform multiple trades per minute, indicating the potential presence of bots or automated systems. Its core feature set enables dapp developers to proactively enforce security measures—such as banning or restricting suspicious addresses—while also providing a clearer view of on-chain trading dynamics.
Beyond security, DeepShield-HFT can serve as a valuable on-chain intelligence tool. By pinpointing which tokens are heavily traded by bots, the model helps users and developers gauge how much of a token’s volume is truly organic. This insight supports more transparent market analyses and informed decision-making.
Like HODL-C1, DeepShield-HFT is trained exclusively on Ethereum on-chain data. Although it can be extended to other networks, the current model focuses on analyzing wallet addresses and their transaction histories within the Ethereum ecosystem. All insights are derived from publicly available blockchain records, ensuring transparency and verifiability.
DeepShield-HFT adopts the same classification framework used in HODL-C1, benefiting from advanced feature engineering and robust model training. For labeling, addresses are considered high-frequency traders if their last 3 out of 50 trades each had a holding period of one minute or less. This strict criterion ensures the model accurately captures addresses engaging in rapid, bot-like trading behaviors.
By leveraging proven machine learning techniques and extensive training data, DeepShield-HFT offers a high degree of reliability in flagging suspicious activity. Its flexible architecture allows for future enhancements, including adaptation to other blockchains and potential real-time detection capabilities.
DeepShield-HFT achieves over 95% accuracy in identifying high-frequency trading addresses, a performance level consistent with our broader suite of security and analytics models. This accuracy metric stems from rigorous training and validation, where out-of-sample testing confirms the model’s ability to generalize effectively to new data.
Because the model borrows directly from the HODL-C1 architecture, it also inherits the proven methodologies that minimize overfitting and ensure stable performance. This gives developers confidence in deploying DeepShield-HFT for mission-critical security applications.
DeepShield-HFT is primarily intended to help dapp developers detect and mitigate bot-driven trading activity. By identifying high-frequency trading addresses, developers can implement tailored security measures such as outright bans, restricted permissions, or clear flagging within user interfaces. This proactive approach preserves the integrity of on-chain markets and protects human users from unfair or manipulative trading practices.
In addition, DeepShield-HFT provides valuable market intelligence for anyone interested in understanding the nature of token trading volume. By highlighting the presence of automated activity, it enables more transparent analyses of market liquidity and trends. As the Web3 ecosystem continues to evolve, DeepShield-HFT stands ready to adapt, offering a reliable foundation for both current security needs and emerging real-time detection scenarios.
DeepShield-FR is our specialized security model designed to detect front-running bots—particularly those engaging in sandwich attacks—on the Ethereum blockchain. Unlike traditional security solutions, DeepShield-FR focuses on early detection by analyzing the first five transactions of each wallet. By identifying transaction signatures that signal potential sandwich attacks, this model estimates the likelihood of an address being used for front-running in the future. With an accuracy rate exceeding 95%, DeepShield-FR empowers dapp developers to take preemptive measures and secure their platforms before threats escalate.
DeepShield-FR offers a unique approach to blockchain security with features tailored for early threat detection:
Early Detection: By focusing on the initial five transactions, the model identifies suspicious behavior patterns before a bot can fully exploit the market.
Signature Analysis: DeepShield-FR scrutinizes transaction signatures to pinpoint indicators of sandwich attacks, distinguishing malicious activity from normal trading behavior.
Actionable Insights: The model provides a probability score for each address, enabling developers to quickly assess risk and implement targeted security measures.
DeepShield-FR is trained exclusively on Ethereum on-chain data. It leverages detailed transaction records, specifically analyzing the first five transactions per wallet. This focused dataset captures the early behavioral patterns and transaction signatures that are critical for detecting potential front-running attacks, ensuring transparency and verifiability through publicly available blockchain records.
DeepShield-FR leverages an XGBoost classification model to detect front-running bots by analyzing the first five transactions of a wallet. Here’s an overview of our rigorous development process:
Data Preparation and Feature Selection: We begin by loading our dataset and extracting a set of predefined features relevant to early transaction behavior. The target variable indicates whether an address exhibits characteristics associated with sandwich attacks.
Model Training and Hyperparameter Tuning: The data is split into training and test sets. We use the XGBoost classifier as our core algorithm. To ensure optimal performance, we optionally perform hyperparameter tuning using GridSearchCV with a 5-fold cross-validation approach, which allows us to fine-tune parameters for the best accuracy.
Model Evaluation: After training, the model's performance is evaluated on the test set. We generate key metrics such as overall accuracy, a detailed classification report, and a confusion matrix. These metrics confirm that our model reliably distinguishes between benign and potentially malicious early transactions.
Feature Importance Analysis: We calculate and analyze feature importances to understand which factors most significantly influence the model’s predictions. This insight helps validate that the chosen features effectively capture the early indicators of front-running behavior.
Model Deployment and Persistence: Once trained and validated, the model is saved to disk for future use. The same model can be loaded and used to predict probabilities on new data, ensuring seamless integration into our security pipeline.
This systematic approach, combining careful feature engineering, hyperparameter optimization, and comprehensive evaluation, ensures high performance when detecting front-running bots on the Ethereum blockchain.
DeepShield-FR consistently achieves an accuracy rate exceeding 95% in detecting front-running bots. Extensive training and out-of-sample testing confirm that the model effectively generalizes to new data, ensuring that early detection remains robust under real-world conditions. This high accuracy not only instills confidence in its deployment but also enables proactive security measures to protect decentralized applications from emerging threats.
DeepShield-FR is an essential tool for dapp developers and security teams looking to fortify their platforms against malicious bot activity. For example, by flagging addresses with early signs of sandwich attacks, the model enables:
Proactive Defense: Implementing immediate countermeasures, such as dynamic address blocking or tailored permission settings, to mitigate potential front-running risks.
Enhanced Security Protocols: Refining smart contract parameters and security policies based on early-warning signals to safeguard user assets and maintain market integrity.
Broader Market Insights: Informing comprehensive threat assessments and security audits, allowing developers to adapt their strategies to the evolving landscape of on-chain activity.
DeepShield-FR thus provides a critical layer of security that not only protects decentralized platforms but also contributes to a safer and more reliable blockchain ecosystem.
The AI MarketSuite is a comprehensive marketplace and development suite built for Web3. It provides a permissionless, self-service interface where users can purchase, deploy, and even create AI models with ease. By uniting a vibrant marketplace with robust development tools, the MarketSuite serves both novice users and advanced data scientists, creating a seamless ecosystem for AI innovation on blockchain platforms.
The features of the marketplace and development suite combine to promote the depth and breadth of available algorithms for Deep3's customers to use across the full range of applications that will exist in Web3.
The marketplace is the central hub where D3L’s suite of AI models is showcased, bought, and sold. Here, users benefit from a highly intuitive, point-and-click interface that simplifies every transaction. All purchases and model deployments are executed through smart contracts, ensuring security and transparency throughout the process.
At the heart of the marketplace is the D3L ecosystem token, which plays a critical role in every transaction. Users pay for models using these tokens, and upon purchase, receive an authentication token that grants access to the associated API or oracle services. This token-driven economy not only streamlines payments but also creates organic demand for D3L tokens, as their utility is directly tied to the seamless functioning of the marketplace. Moreover, tokens can be used to access premium features, provide royalties for third-party developers in future iterations, and even serve as collateral for model submissions, reinforcing a sustainable, incentivized ecosystem.
The marketplace is designed to cater to the diverse needs of its users. Comprehensive implementation tutorials, a straightforward rating system, and integrated support channels ensure that every customer—from a seasoned developer to a newcomer—can confidently navigate the platform. As the most highly trafficked property in the D3L ecosystem, the marketplace also acts as a central repository for research, training materials, and community insights, establishing it as the nexus of Web3 AI innovation.
Complementing the marketplace is a robust development suite that empowers users to create and customize their own AI models. This suite is designed with flexibility in mind, offering tools that range from no-code interfaces for beginners to advanced development environments for experienced data scientists.
The development suite covers the entire machine learning pipeline:
This suite is particularly valuable for third-party data scientists looking to experiment with crypto data. Extracting, transforming, and loading (ETL) crypto data is notoriously time-consuming and expensive due to the sheer volume of data and the complex, unstructured nature of blockchain nodes. The Deep3 platform streamlines this process by eliminating roughly 90% of the time required to build a model end to end. This reduction in effort enables ML experts from other fields to experiment with Web3 ML during their spare time—on nights and weekends—thus fostering innovation in crypto. Moreover, as the development suite is natively integrated with the marketplace, there is a clear path to monetizing their models, which further incentivizes experimentation and drives the overall growth of the ecosystem.
A standout feature of the development suite is its AutoML functionality. This tool automates key parts of the machine learning process—such as model training and inference—allowing users with limited technical expertise to generate effective models. With guided parameter selection, smart validation recommendations, and rapid deployment options, AutoML lowers the barrier to entry while maintaining high standards of performance.
The ultimate vision for the AI MarketSuite is to create an open, dynamic ecosystem that accelerates AI innovation in the decentralized world. Its core objectives include:
In its final form, the AI MarketSuite will not only be a marketplace for AI models—it will be the heartbeat of a decentralized AI economy, enabling continuous innovation, efficient model development, and sustainable growth across the Web3 landscape.
The D3L DAO is the cornerstone of our AI governance, dedicated exclusively to managing and guiding all machine learning activities across the Deep3 Labs ecosystem. From the very first model update to ongoing enhancements in data pipelines and token economics, the DAO ensures that every decision related to AI development is made transparently and with community involvement. Rather than ceding full control over the entire business, we have chosen a focused approach—starting with the AI/ML operations—to build trust, share value, and gradually transition more control to the community as our collective expertise grows.
The primary function of the D3L DAO is to democratize the AI development process. It does so by allowing community members to review, validate, and vote on key proposals regarding new model implementations, data integrations, and enhancements to our technical infrastructure. For instance, when Deep3 Labs updates a model, the new version is not pushed live automatically; instead, our system listens for a passing governance vote, and only then is the updated model deployed via the API. This rigorous, community-driven process ensures that all significant AI-related changes are subjected to collective scrutiny, thereby preserving the integrity of our AI products while protecting proprietary trade secrets.
The DAO’s mandate is intentionally narrow at inception, covering decisions directly related to AI/ML activities. However, its scope will gradually expand to include strategic treasury functions such as setting fees, determining royalties, and managing token emissions—all critical components that align with our mission to share the value generated by our technology.
The D3L DAO operates on a robust governance framework powered by a dedicated governance token. This token is not only essential for casting votes but also serves as an incentive mechanism to reward active and valuable participation. To prevent voting power from being overly concentrated among wealthy stakeholders, we are exploring advanced mechanisms such as quadratic voting, ensuring a fair and balanced decision-making process.
The governance process is designed to be seamless and automated. Once a proposal receives the requisite support from the community, our smart contract system automatically triggers updates, such as pushing new model data to the API. This integration of automated workflows with decentralized decision-making underscores the DAO’s critical role in maintaining the agility and responsiveness of our platform.
The strategic objectives of the D3L DAO are multifaceted:
Transparency and Accountability: By opening up key aspects of the AI/ML pipeline to community oversight, the DAO ensures that model development, performance metrics, and data integration practices are transparent and accountable.
User Empowerment: The DAO provides users with a direct pathway to influence how AI models are built and deployed, thereby enhancing user agency in an industry where opaque processes are the norm.
Sustainable Value Sharing: Through governance over fee structures, royalties, and token emissions, the DAO ensures that the value generated by our AI models is equitably shared among all ecosystem participants.
Risk Management: The incremental approach to ceding control helps mitigate risks by balancing innovative decentralization with the need to protect sensitive operational details and trade secrets.
Continuous Improvement: The DAO’s iterative review process, including automated updates based on governance votes, ensures that the Deep3 Labs platform remains at the cutting edge of AI/ML technology.
To illustrate the DAO’s scope and decision-making process, here are some examples of the types of proposals that may be brought forward for community consideration:
While the initial focus of the D3L DAO is strictly on AI/ML activities, its role is expected to evolve significantly as our community matures and as the broader ecosystem gains a deeper understanding of decentralized governance in AI. In the future, we envision the DAO taking on additional responsibilities such as overseeing broader operational decisions, updating the token economics, and even engaging in strategic partnerships. As these changes are implemented incrementally, each expansion will be carefully managed to ensure that the benefits of decentralization are realized without compromising the operational integrity or security of the platform.
In summary, the D3L DAO is not just a governance tool—it is a foundational element of our mission to democratize AI. By giving users a direct voice in how models are built, validated, and deployed, the DAO transforms passive consumers into active participants in shaping the future of AI/ML. This pioneering approach not only enhances transparency and accountability but also drives sustainable growth by ensuring that the value generated by our technology is shared equitably across the entire ecosystem.
User Empowerment: Delivering a user-friendly platform that makes advanced AI technology accessible to everyone, regardless of technical skill level.
Seamless Integration: Providing robust, end-to-end development tools that allow users to create, test, and deploy models quickly and securely.
Token-Driven Economy: Leveraging D3L ecosystem tokens to facilitate transactions, incentivize contributions, and drive organic demand across the platform.
Community and Collaboration: Serving as a central hub for research, training, and community engagement that brings together data scientists, developers, and industry leaders.
Scalability and Adaptability: Designing the platform to evolve over time with additional features, such as third-party model submissions, integrated DAO oversight, and advanced AutoML capabilities.
Hōkū, by Deep3 Labs, is a proof-of-concept decentralized application that leverages AI/ML to demonstrate the potential of applying AI-powered recommendation systems to the crypto token traading space (similar to those found in Netflix, Tiktok, YouTube, etc).
You can find Hōkū at:
Accretion, by Deep3 Labs, is a proof-of-concept decentralized application that leverages AI/ML to demonstrate new blockchain advertising and ad targeting capabilities.
You can find Accretion at:
Deep3 Labs integrates advanced AI into blockchain, transforming how users interact with Web3. Our dapps in trading, advertising, and security deliver personalized insights, precision targeting, and robust protection against bots. By leveraging on-chain data and sophisticated algorithms, our trading dapps provide actionable market intelligence, while our advertising tools drive tailored engagement and our security solutions detect suspicious activity. We build these dapps not only to deliver real value but also to showcase our ML tools—which are available for other developers—to drive a smarter, safer, and more rewarding blockchain ecosystem.
Getting Started
You do not need to connect your wallet to begin exploring Hōkū. However, connecting your wallet unlocks some personalization features. To connect your wallet, click "Connect Wallet" in the upper right corner.
We currently support MetaMask, Coinbase and Phatom wallets, as well as any Wallet Connect compatable wallet.
Similarly, you do not need to sign in order to explore Hōkū. However signing in unlocks additional features such as Favorite Wallet lists. To sign in, click "Sign in" in the upper right corner.
When signing in, you will be requested to sign a message with your wallet. This proves that you are the owner of that wallet and does not grant Deep3 Labs any control or privileges over your wallet or the tokens you hold in it. There are no transaction fees associated with this operation.
Once complete, these two controls will show that you are fully connected and signed in.
The primary navigation controls are found on the left-side of each page. Home is the main landing page for the app. My Dashboard contains a variety of information about the wallet you've connected with. Cluster Explorer is our interactive, 3D network explorer, which shows all of the active traders on the chain. Cluster Trends displays real-time buying and selling trends within each trading cluster. My Favorited List shows a list of wallets that you've marked for tracking. Notifications shows the active and past alerts for trading trends that have emerged from our real-time trading cluster data feed. Settings and Terms & Privacy contain basic information about Hōkū.
Within each page, you may find additional navigational controls and typically those will be found at the top of the page. For most pages, these consist of clearly titled drop-down filters or buttons.
On the Cluster Explorer specifically you'll find a series of buttons across the top.
From right to left, these buttons are:
Within the DeepLeap menu, you'll find a familiar AI-chat interface and controls to search through the lists of addresses that it discovers (additional information on using DeepLeap is below).
Once you've executed a search, use the controls above the chat interface. They will allow you to (from left to right): hide the interface, return to the first wallet in the list, go back one wallet in the list and go forward to the next wallet in the list.
When you've selected a wallet within the cluster explorer, an informational pane will open in the bottom right corner of the window.
Here you can toggle between viewing that wallets balances or its past trades.
You can also view its historical balances from 7 days ago, 14 days ago and 21 days ago.
Above the wallet info pane, you'll find information on the cluster that the wallet is found in.
This can also be expanded to show additional information about that cluster.
The real-time trends shown on the cluster trends page can be filtered by (from left to right) the type of transaction, all network activity vs DEX only, with or without stablecoins and over two time horizons.
Additional information about the traders that are included in the cluster can be viewed by clicking the info icon at the top of each cluster card.
The cluster cards themselves can also be re-arragned by clicking and dragging or removed from view and then added back later. If you click the "X" on any cluster card, you can return it to view by clicking "Add cluster"
The Cluster Explorer is a cornerstone feature of Hōkū that transforms complex blockchain data into an interactive, visually intuitive experience. By leveraging advanced machine learning and clustering algorithms, it groups wallets based on similar trading behaviors and transaction dynamics. This not only makes the data more accessible but also empowers you to identify trends and potential opportunities that might be missed with conventional blockchain explorers.
Imagine being able to ask your blockchain explorer in plain English—no complex queries or coding required. With DeepLeap, you can simply type commands like "find wallets that made 100x buying memes early" and instantly uncover the precise insights you need. This natural language search capability makes data exploration intuitive and accessible, allowing you to focus on what matters: spotting trends and seizing opportunities.
Hōkū’s real-time cluster trends give you market data as it happens—even faster than traditional explorers like Etherscan. The tool constantly watches different groups of traders, showing you what people are buying or selling right now. This early view lets you see trends before they spread across the market and lead to big pumps or dumps. With these early signals, you can react quickly and take advantage of new opportunities before the rest of the market catches on. This powerful feature makes it easier for you to make smart trading decisions and stay ahead in a fast-moving market.
Hōkū’s real-time notifications ensure you never miss a market-shifting event, even when you're working in another browser tab. With alerts that have already captured winning trades like 208x on FAI, 183x on AIXBT, 180x on MINIDOGE, 120x on TITCOIN, 39x on AION, 22x on REI, 20x on DISTRIBUTE, 13x on AWS, 12x on VIRTUALS, 8x on ANON, 5x on LION, and 4x on both VVV and MINT, this feature provides an immediate pulse on the market's most lucrative moves. You can toggle system notifications on to receive desktop or mobile alerts, ensuring you’re always connected and ready to act on high-impact trading signals, no matter where you are.
Hōkū’s proprietary short-term price predictions are based on a series of 18 different machine learning models trained on a history of real-time notifications issued in the app. In out-of-sample testing, the false positive rate of these models is often under 1%. They are displayed alongside every notification in the app in three different places:
On pop-up notifications while the cluster explorer is being used, where the color of the dice represents the magnitude of the predicted probability of the stated price change
On each notification in the active notification screen, where the number of cells in the price change badge represents the magnitude of the predicted probability of the stated price change.
In the token details modal in a complete tabular format which shows all time-periods and price thresholds for which a prediction is available.
The complete tabular format appears as follows:
On the active notifications screen, you can select which time periods and price thresholds are shown for the buy and sell notifications separately. Choosing "Auto Select" will display the most actionable time period and price threshold values for each notification automatically.
Based on the data we've seen so far, this one is actually quite simple. Turn on system notifications (so you don't miss an alert when you're surfing Web3.
The keep an eye out for trend notifications! The more clusters the token is trending in, the stronger the signal. But make sure to catch it early, and DYOR on all standard security checks (we're actively working on in-app security checks).
What makes Hōkū so powerful when it comes to identifying these trends early is our ability to drill-down into the broader market to identify trends within individual trading clusters. It's far more effective than the typical market wide "hotlists" because these market-wide trends have to start somewhere. Hōkū allows you to send the trend as it forms within one group of traders, then spreads to the rest of the market, giving you the ability to get in before the rest of the market does.
Our AI-powered natural language search is the most powerful blockchain search utilities in the industry, especially when it comes to finding other traders to learn from. Instead of endlessly clicking through transactions on Etherscan, you can simply describe the kind of trader you'd like to learn from, like expert meme token traders.
From there, you can easily look through what those traders are buying, how long they're holding their investments and how they're taking profits.
We even include our own proprietary machine learning insights to help you understand each trader, such as the HODL-C1 model and our Win Rate prediction. This puts powerful insights, that would otherwise take hours to glean from other platforms, at your fingertips relating to how long a trader holds their investments and how often they make a profit.
For any questions or feedback on the app, reach out to us via one of our community platforms or send us an email at hello@deep3.ai.
Our API provides seamless access to all Deep3 Labs AI models, enabling developers to integrate advanced machine learning capabilities directly into their dApps. Initially, our API is open and free to use—you simply sign in with your Ethereum wallet to generate an API key.
Looking ahead, access to the API will be gated by our utility token, ensuring that users engage with the platform in a way that reflects its value and fosters a sustainable ecosystem. Additionally, future products may offer on-chain verification for the data retrieved through our API, either by leveraging external providers like Truebit or API3, or by building our own oracle solutions.
For more details, complete documentation, and to get started with our API, please visit the Deep3 Labs Developer Portal
Reposition camera to center.
Search for a wallet with its address hash.
Locate the connected wallet in the cluster explorer.
Open DeepLeap, our AI-powered network search.
Clicking the left or right buttons next to the cluster label automatically move the camera to the view the previous or next cluster in the explorer.
You do not need to log in to begin using Accretion. All targeting interfaces are accessible immediately. However, to utilize the monitoring capabilities in Accretion, you will need to create an account and login.
You can create an account using Google, Ethereum or your email address.
Once you've created an account and logged in, you will see your profile icon in the upper right corner. To log out, simply click your profile icon and select "Log out".
There are two main navigational schemas: a top menu and a side menu.
Along the top, you can select which stage of the advertising process you wish to operate on. Targeting allows you to use our machine learning models to target those wallets that would be of the highest value for your campaign. Monitoring allows you to track which of your targeted wallets "converted", meaning those wallets that interacted with your smart contract.
The left-hand navigation menu allows you to select one of the machine learning models that Deep3 Labs has developed for ad targeting. HODL-C1 is a DeFi model that identifies long-term holders. StakeSage-L is a model designed specifically for Liquid Staking Derivative platforms as it predicts how much Ethereum a given address will stake over its lifetime.
Accretion allows you to first export a list of addresses to use in a campaign and then monitor which of those addresses interacted with your smart contract during or after your campaign.
To develop a targeting list and export those addresses, you can use any combination of filters and drop-down menus in order to arrive at a suitable address list.
With the appropriate filters in place, you can then select all or select individual wallets, then click "Export" to download a CSV file of the addresses you wish to target in your campaign.
If you want to monitor which of the targeted addresses convert, start by clicking "Monitor". This will take you through a series of steps to create a custom monitoring agent.
The first step in this process is selecting whether you wish to monitor interactions with a token or a different type of smart contract. In this example, we'll select "Smart Contract."
You'll then be prompted to enter the smart contract you wish to monitor, and optionally, a value threshold associated with what you would define as a "conversion."
And finally, you can provide a name for this monitoring agent and set a date window for the start date and end date of your campaign.
When you're done, click "Finish."
To view the results of your campaign, click the "Monitoring" menu item at the top, then click the campaign you would like to inspect.
The details associated with the wallets you targeted in your campaign will then be displayed.
Say you operate a liquid staking platform and you want to target new users, but you want to ensure they are likely to stake at least 10 ETH over their lifetime. These analytical capabilities can be critical to maintaining a profitable return on your marketing investment given your known customer aquisition cost.
First, you could start by filtering on addresses that have a lifetime staking value of 10 ETH or more.
Next, to ensure that the addresses you choose are in fact potential new business for you, you can select a competitor from the "First Platform" list.
And finally, to maximize the conversion potential among the addresses you target, you could select low loyalty scores. Loyalty score measures the affinity an address has for a particular staking platform. High loyalty scores typically only use one platform.
You've now created an ideal targeting list for your campaign. Click "Export" to download your address list. To track your campaign, click "Monitor" and follow the steps outlined above.
For any questions or feedback on the app, reach out to us via one of our community platforms or send us an email at hello@deep3.ai.
The blockchain needs AI for better UX and AI needs blockchain for better governance.
To attract mainstream users, Web3 platforms must deliver mainstream user experiences and that means creating experiences that are more comparable to those in Web2, where ML-driven personalization and optimization have become the norm.
Machine learning transformed the internet by enabling websites to deliver personalized content, making it easier and faster to find relevant, engaging information. This innovation brought more people online, creating a dynamic, user-friendly space that anticipates and meets individual needs.
It works by analyzing vast amounts of user data to train algorithms that predict what users would do or want, optimizing search results, tailoring advertisements, and suggesting relevant products or content. But, that ultimately means you had an important hand in all of this: without the user data you generated, these algorithms couldn't exist.
Today's tech titans have discovered the most lucrative virtuous cycle of our time. Using machine learning to create personalized experiences attracts more users that generate more data allowing them to create even more value that attracts new users. This cycle allowed companies like Google, Facebook, and Amazon to dominate, making it hard for new competitors to succeed.
Products like Google Search or Facebook are "free", because companies use ML to turn your data into their profit.
Just like too much of a good thing can be bad, our reliance on ML has also worsened some of the most complex problems of our day. Algorithms can reinforce biases, worsening things like racial discrimination in hiring and lending. We also now know that personalized content contributes to mental health issues, such as depression or anxiety, by creating echo chambers and promoting negative or harmful content.
What's worse, even though its your data that allows machine learning platforms to function, there's very little you can do to control how these algorithms affect you.
At Deep3, we envision a future where artificial intelligence is seamlessly integrated into Web3—empowering users and developers alike. By leveraging the unique strengths of decentralized networks, our platform paves the way for a more transparent, secure, and intelligent digital ecosystem. This isn’t just about technology; it’s about rethinking how data and value are shared in a user-first internet.
Our platform is built with simplicity in mind. Whether you’re a blockchain developer or a curious user, Deep3’s intuitive interfaces let you harness the power of machine learning without getting bogged down by complex code. With no-code dapps for everyday users and low-code APIs for developers, exploring AI on Web3 has never been more straightforward.
Deep3 is committed to breaking down the barriers to advanced AI tools. We’re making it possible for anyone, regardless of technical expertise or resources, to participate in and benefit from AI innovation. This inclusivity ensures that the opportunities and rewards of machine learning are shared widely, sparking creativity and driving new ideas across the community.
In the spirit of decentralization, Deep3 puts control back into the hands of its users. Our platform is designed so that every participant can have a say in how AI evolves within Web3. By empowering the community to govern the system, we ensure that the development and use of AI tools are transparent, fair, and aligned with the collective interests of all users.
Exos, by Deep3 Labs, is a proof-of-concept decentralized application that leverages AI/ML to demonstrate new blockchain security capabilities.
You can find Exos at:
Exos is Deep3 Labs’ proof-of-concept no-code blockchain security platform that harnesses advanced AI to proactively protect decentralized applications. By leveraging cutting-edge machine learning models, Exos delivers predictive security solutions that help safeguard blockchain ecosystems from disruptive bots and malicious actors. With model accuracies regularly exceeding 95% and training based on billions of transactions, Exos transforms how security is managed on-chain.
Exos integrates two equally powerful models—DeepShield-HFT and DeepShield-FR—to offer comprehensive protection across blockchain networks.
DeepShield-HFT detects high-frequency trading addresses that are predominantly bot-controlled, flagging suspicious trading activity with minute-level precision.
DeepShield-FR targets front-running threat actors, identifying addresses likely to engage in sandwich attacks before they escalate into significant risks.
Together, these models provide a dual-layered defense that not only monitors rapid trading behaviors but also anticipates potential front-running threats. Exos enables configurable deny lists and throttling for DEXs, launchpads, and other dapps, allowing developers to dynamically mitigate bot-related risks with just a few clicks.
Exos is built on Deep3 Labs’ state-of-the-art machine learning infrastructure. Our models are trained on vast amounts of on-chain data—billions of transactions across networks like Ethereum and Base—to ensure robust and accurate threat detection. The platform leverages proprietary algorithms that continuously update and refine predictive security metrics, providing users with high-resolution insights into the bot versus human-controlled dynamics within their dapps.
Exos is designed to serve a variety of practical applications:
Protecting Core Users: By detecting and mitigating bot activity, Exos helps preserve the integrity of decentralized applications, ensuring that genuine users remain loyal customers.
Enhancing Brand Reputation: With proactive bot mitigation, brands can avoid negative experiences often caused by malicious automated trading, fostering trust and long-term engagement.
Improving Market Analytics: The platform’s high-resolution user versus bot analytics provide invaluable insights for dapp developers, allowing them to fine-tune their security protocols and optimize smart contract performance.
These capabilities not only safeguard individual dapps but also contribute to a healthier, more resilient Web3 ecosystem overall.
Exos is poised to evolve alongside the rapidly advancing blockchain security landscape. Our planned enhancements include:
Real-Time Detection: Transitioning from historical analysis to real-time monitoring to instantly flag malicious activity as it occurs.
Additional Security Models: Expanding our portfolio to include models that analyze contract owner behaviors and other critical security aspects.
Network Expansion: Extending our predictive security solutions to additional blockchain networks, ensuring broader protection across the decentralized landscape.
These forward-looking improvements are designed to continually enhance the platform’s effectiveness and keep pace with emerging threats in the blockchain space.
Exos represents the next generation of blockchain security by combining advanced predictive AI with a user-friendly, no-code interface. By integrating robust models that identify both high-frequency trading bots and front-running threat actors, Exos empowers dapp developers to proactively secure their platforms while delivering a superior user experience. As we continue to innovate—with plans for real-time detection, additional security models, and expanded network support—Exos remains at the cutting edge of predictive blockchain security, making it an essential tool for safeguarding the future of Web3.
Hōkū is the first 3D blockchain explorer powered by advanced machine learning, designed exclusively for the trading world. By blending immersive visualization with smart analytics, Hōkū delivers a Web2-like user experience on Web3. It empowers traders to uncover hidden patterns, navigate millions of wallets, and capitalize on actionable insights, transforming the way blockchain data is explored and monetized.
Even in this initial version, Hōkū offers crypto traders a series of ground-breaking features provide a trading edge that no other crypto dapps can offer.
Hōkū's stunning 3D interface lets you effortlessly navigate through millions of wallets, revealing complex relationships and hidden trends across billions of trades. This immersive experience makes blockchain data tangible and accessible like never before.
Stay ahead of the market with cluster-specific alerts. Hōkū's advanced ML algorithms monitor token momentum and deliver real-time notifications—case studies have even demonstrated profit potentials as high as 208x.
Each time a notification is issued, a series of 18 machine learning models are run to predict the likelihood of various price increases or decreases over a 1-hour, 4-hour and 24-hour time period. You'll the option to select which time period or price threshold you wish to view.
Unlock unique wallet insights with our intelligent search tool. Whether you're looking for addresses that achieved 100x gains trading memes or early investors in promising tokens like VIRTUALS, Hoku’s AI-driven search capabilities provide unparalleled precision.
Gain instant visibility into live trading dynamics. Hōkū aggregates real-time data across trading clusters, offering insights that are faster and more detailed than traditional explorers like Basescan or Etherscan.
Hōkū's platform is built on three key technological pillars that together provide a powerful, real-time trading experience. By integrating high-speed data feeds, advanced clustering algorithms, and comprehensive trading analytics, Hōkū offers traders an edge in navigating the dynamic blockchain landscape.
Hōkū features a state-of-the-art, real-time data feed that delivers on-chain information faster than even the major block explorers. This rapid delivery of data ensures that traders receive the most up-to-date insights and market activity, enabling timely decision-making and immediate response to fast-moving trends.
At the heart of Hōkū's analytical capabilities lies our advanced clustering model. This innovative algorithm groups wallets based on risk tolerance and transaction dynamics by generating sophisticated embeddings that capture intrinsic wallet characteristics. Through dimensionality reduction, these high-dimensional data sets are transformed into a three-dimensional manifold that powers Hōkū's immersive 3D explorer interface, making complex trading relationships visually intuitive and easily navigable.
Complementing the data feed and clustering model, Hōkū integrates a robust suite of trading performance metrics and predictive analytics. By processing billions of transactions, our platform delivers precise predictions and actionable insights that help traders optimize their strategies. These analytics not only reveal real-time market trends but also forecast potential profit opportunities, empowering users to trade with greater confidence and precision.
Hōkū is designed to transform trading by providing actionable insights that directly help traders maximize their profits. Our platform’s smart notifications and AI-powered analytics have already led to remarkable potential returns—traders could have seen profit multipliers of 208x on FAI, 183x on AIXBT, 180x on MINIDOGE, 39x on AION, 22x on REI, 13x on AWS, and 12x on VIRTUALS. By delivering real-time, cluster-specific alerts and detailed wallet analyses, Hōkū empowers traders to identify lucrative opportunities early and capitalize on market trends before they become widely recognized. This direct impact on trading performance makes Hoku an indispensable tool for anyone looking to gain a competitive edge in the dynamic world of blockchain trading.
Looking forward, Hōkū's evolution centers on the development of automated trading systems. We plan to create increasingly sophisticated trading agents that learn from real-time market activity and adapt by following strategies of selected successful traders. These intelligent agents will be designed to execute trades autonomously, further enhancing the platform’s capability to deliver value and drive user engagement. Additional enhancements include integrating more data sources, refining our ML models, and extending the platform’s features to support a broader range of trading strategies.
Hōkū is not just a trading tool—it’s a revolutionary platform that reimagines how traders interact with blockchain data. By offering immersive 3D visualization, smart notifications, AI-powered wallet search, and real-time trend analysis, Hōkū provides a comprehensive suite of features that empower traders to make informed decisions. As we continue to innovate with automated trading agents and expand our data integration capabilities, Hōkū is poised to lead the way in AI-powered trading on Web3.
Our Bulk Downloader utility allows users to retrieve a batch of predictions from one of our models for a given address list.
You can find the Bulk Downloader at:
Our ecosystem is built around a dual-token model that not only facilitates seamless transactions and access to our cutting-edge AI models but also gives users a meaningful stake in our platform’s success.
At its core, our ecosystem comprises two distinct tokens: the governance token (gD3L) and the utility token (uD3L). The governance token enables community participation in key decisions—ranging from model updates to economic policies—while the utility token powers day-to-day transactions within the marketplace, from purchasing AI models to collateralizing submissions. Complementing these tokens is our innovative staking program, which rewards active participants and ensures that real contributors are recognized in our rapidly evolving digital landscape. Graphically, the ecosystem can be summarized as follows:
Together, all of these components create a dynamic, self-sustaining economy that not only drives the adoption of AI on Web3 but also puts control and value directly in the hands of its users.
Accretion is a groundbreaking advertising platform that harnesses behavioral prediction algorithms to revolutionize ad targeting on Web3. By leveraging advanced ML models, Accretion enables businesses to precisely identify, target, and engage valuable users, merging the best practices of Web2 advertising with the transparency and efficiency of blockchain technology. Accretion is the first Web3 platform that offers the same type of state-of-the-art targeting capabilities that are used by the most sophisticated Web2 advertisters. Our models directly predict which addresses will engage in the behaviors that constitute key business outcomes, such as a closed sale or a new customer aquisition.
Accretion empowers businesses to target specific user segments using outputs from our state-of-the-art DeFi models. These models analyze on-chain behavior to predict user value and acquisition potential. The platform not only facilitates targeted advertising but also tracks subsequent interactions with smart contracts, allowing advertisers to measure the real-world impact of their campaigns. Additionally, Accretion supports campaign optimization, ensuring that ad strategies continuously improve to meet key performance indicators.
Built on robust machine learning algorithms and deep blockchain analytics, Accretion leverages vast amounts of on-chain data to generate actionable behavioral insights. Our proprietary models, developed from years of expertise in both AI and blockchain technology, power the platform’s precision targeting and real-time tracking capabilities. This technological foundation enables Accretion to deliver ad targeting that is both highly accurate and adaptable to dynamic market conditions.
Accretion empowers businesses to deploy highly targeted advertising campaigns by leveraging advanced predictive models, ensuring that every marketing dollar is spent efficiently. For example, companies planning token airdrops can integrate HODL-C1 to target long-term holders, thereby enhancing token stability and fostering community loyalty. Likewise, liquid staking platforms can use StakeSage-C to identify potential new customers who are most likely to engage in liquid staking for the first time, while StakeSage-L helps target high-value users who are predisposed to commit larger staking amounts. These model-driven insights translate into more effective ad spend, higher engagement rates, and improved campaign performance, ultimately driving tangible growth in the decentralized ecosystem.
Looking ahead, Accretion will expand its capabilities by enabling ad creation and delivery directly within the platform. We plan to integrate with emerging Web3 display networks and leverage our sister platform, Hoku, as a channel for delivering targeted ads. Future enhancements will also include sophisticated campaign optimization tools that adapt to evolving market trends, ensuring advertisers can consistently achieve superior targeting and performance outcomes.
Accretion is redefining digital advertising on Web3 by combining the precision of AI-powered behavioral insights with the efficiency of blockchain technology. By bridging the gap between proven Web2 methodologies and the innovative possibilities of Web3, Accretion offers businesses a powerful tool to drive user acquisition and optimize campaign performance. Explore Accretion today to see how you can transform your advertising strategy and unlock new growth opportunities in the decentralized era.
To use Exos, you must first connect a wallet.
We currently support Metamask, Coinbase wallet and any Wallet Connect compatabible wallet.
Once connected, you'll be shown the main user dashboard.
The dashboard allows you to toggle between supported networks (currently, only Ethereum is available) and shows the number of times your wallet has suffered a front-run attack.
On the left-hand side, you'll find menu items for the other two main sections of the app. Each uses one of Deep3 Labs' security models to provide information and address exports from the DeepShield-HFT model and the DeepShield-FR model.
Within each of our model-specific pages, there will be up to three sub-menus. Overview provides network-level aggregate information about the particular bot type. Explore allows you to view all addresses on the networks suspected to be a bot. Export provides an interface to download a csv file containing a list of addresses suspected to be a bot.
The primary feature of Exos is its ability to extract user-configurable lists of addresses that are suspected to be a bot.
For front-running, or sandwich attack, bots, you can configure how recently that bot has been seen operating on the network.
For high-frequency trading bots, you can configure the model's "sensitivity." The higher the sensitivity value you select, the more certain you can be that the resulting addresses are in-fact machine-controlled (i.e, bot) addresses.
Say you've built a platform that allows users to claim rewards, but you want to prevent machine-controlled addresses from interacting with your contract. You can easily configure and download this list of addresses using Exos.
In this case, you're likely to select a mid-level of sensitivity, so that you can reduce the likelihood that a legitmate user would be blocked from claiming a reward.
You can also use the dynamic plots to guide your selection. There is typically a clear "elbow point" in each curve. Beyond this point, it becomes exceedingly unlikely that an address would be misidentified by the model.
You can manually inspect the transaction histories of identified addresses in the list below.
Once you're satisfied with the configuration, simply click "Download".
If you operate a DEX, front-running bots can harm novice traders, making it harder to retain and grow a trader base. Our model identifies these addresses before they've attacked dozens or even thousands of users and Exos makes it simple to identify and block them.
Just like the HFT interface, you can configure your address list then download a CSV file.
Our unique 3D explorer allows you to inspect the bots that are active at a given time. By using machine learning techniques, we're able to create a 3D map of these addresses so that similar bots appear near one another in the space.
Once you click an address inside the visualizer, you'll find a comprehnsive list of its attacks below.
Additional transaction information can then be viewed on Etherscan by clicking the transaction hash links.
For any questions or feedback on the app, reach out to us via one of our community platforms or send us an email at hello@deep3.ai.
Deep3 Labs’ legal strategy is a cornerstone of our operational and innovation framework. From our early days, we have prioritized legal excellence to protect our intellectual property, ensure regulatory compliance, and set a robust foundation for future growth. Our proactive approach to legal matters has not only safeguarded our technology but also positioned us favorably for future mergers, acquisitions, and market expansion.
We have maintained an in-house general counsel since our inception, ensuring that every aspect of our corporate structuring, agreements, and non-disclosure arrangements meets the highest legal standards. Recognizing the critical importance of intellectual property in our industry, we engaged Dickinson Wright—a top-tier intellectual property law firm—to file comprehensive patents covering the key components of our platform. These patents, pursued in the United States and internationally, have proven prescient; for example, our early "AI agent marketplace" patents anticipated the recent boom in Web3 AI and agent marketplaces.
Deep3 Labs is committed to operating within the highest standards of regulatory compliance. Our token presale was conducted in full accordance with SEC guidelines, a rare achievement among crypto startups at our stage. By aligning our practices with strict U.S. securities laws, we have built a foundation of trust and transparency, setting us apart from many industry peers.
We have intentionally chosen to domicile our company and intellectual property in the United States. Despite the additional challenges this entails, the U.S. remains the best jurisdiction for protecting intellectual property and preparing for potential mergers and acquisitions. Our partnership with Bull Law—a leading U.S. securities law firm—has been instrumental in guiding us through the SEC filing process and ensuring ongoing regulatory compliance.
In addition to our corporate and IP strategies, we are establishing a solid framework for the management of our token ecosystem. The foundation and token treasury are being set up in collaboration with Cambpells, one of the top firms in the industry. Operational management of the foundation will be handled by Lemma Solutions, a leading service provider with an impressive track record working with Arbitrum, ENS, Kava, and other prominent crypto firms. This strategic partnership ensures that our token treasury is managed with the highest level of expertise, supporting both the financial stability and long-term growth of our ecosystem.
Deep3 Labs’ comprehensive legal strategy is a testament to our unwavering commitment to excellence, transparency, and long-term growth. By combining robust intellectual property protection, stringent regulatory compliance, and strategic partnerships with industry-leading law and treasury management firms, we have built a foundation that not only secures our innovative technology but also positions us for future market success. This meticulous approach to legal and corporate governance makes Deep3 uniquely attractive to investors seeking a secure, compliant, and forward-thinking venture, and to users who want to participate in a truly decentralized, trustworthy ecosystem that empowers and rewards its community.
The Deep3 Labs utility token is tentatively named "uD3L".
The uD3L token serves as the de facto payment medium for purchasing access to AI products, executing transactions within our marketplace, and even acting as vital collateral for model submissions. In essence, uD3L is what powers the day-to-day activities on our platform, ensuring smooth, efficient interactions between users, developers, and the underlying technology.
A detailed breakdown of our token economics will be shared in the near future. What we can confirm now is that the uD3L token will have a variable supply that is controlled by the governance framework. This means that adjustments to the supply will be carefully managed by the community through the D3L DAO, ensuring that token availability remains aligned with market needs and platform growth. Furthermore, the only mechanism by which new utility tokens are minted is through our innovative staking program, which rewards active participation and helps maintain a balanced and sustainable ecosystem.
The uD3L token plays a multifaceted role within our ecosystem. It is used for:
Transaction Settlements: Every purchase or service within the D3L marketplace is executed using uD3L tokens, ensuring fast, secure, and efficient payments.
Collateral and Incentives: For data scientists listing models for sale, uD3L tokens serve as vital collateral, aligning incentives between buyers and sellers and ensuring model quality and accountability. They will also receive uD3L when their models are used.
Ecosystem Engagement: As the primary medium for daily interactions on our platform, uD3L drives usage and engagement. In future iterations, the token may also be used to unlock premium features or be integrated into reward schemes.
By providing a clear, consistent medium for transactions and incentives, the uD3L token is essential for the operational integrity and growth of the Deep3 Labs ecosystem.
Governance token holders receive utility tokens from our unique staking program.
The Deep3 Labs staking program is a critical pillar of our token ecosystem, designed to seamlessly link our governance token (gD3L) with the utility token (uD3L). By staking gD3L tokens, users earn newly minted uD3L tokens, creating a powerful incentive to participate in our decentralized network. This program not only rewards long-term commitment but also stabilizes the utility token’s market value and reinforces overall platform growth. The staking process is governed by transparent, on-chain rules and is subject to ongoing adjustments through DAO oversight.
For our initial implementation, we have selected a linear approach to determine the staking yield. In this model, the yield—the number of uD3L tokens earned per staked gD3L—gradually increases over time. This time-varying yield creates a strong disincentive to unstake, as the longer your tokens remain staked, the more rewards you accumulate.
The linear formula is defined as follows:
Where:
t is the number of seconds since staking rewards began accruing.
Ymin is the initial issuance rate (e.g., 1 uD3L per gD3L per year).
Ymax is the target issuance rate to be reached after a predetermined period.
Tmin is the initial period during which no rewards are earned.
Tmax is the number of seconds until which the yield reaches Ymax.
For time periods between Tmin and Tmax, the yield increases linearly from Ymin to Ymax. After Tmax, the yield remains constant at Ymax. This design ensures that stakers are rewarded more as their staking duration increases, aligning individual incentives with the long-term health of the ecosystem.
In addition to the linear approach, we have developed non-linear staking models—such as a logistic S-curve or an exponential explosive curve—that could be deployed in future iterations. These models would provide a different reward dynamic, potentially offering steeper yield increases or more controlled emission growth under certain market conditions. Importantly, our design ensures that if a new staking contract is adopted, users who upgrade will not lose their accumulated staking maturity, preserving the rewards they’ve earned over time.
The non-linear formula is defined as follows:
Where:
t is the number of seconds since staking began.
Ymin is the initial issuance rate (e.g., 1 uD3L per gD3L per year).
Ymax is the target issuance rate to be reached after a predetermined period.
k is a steepness parameter the controls how quickly the curve transitions from Ymin to Ymax.
We ultimately chose the linear option due to its simplicity. However there are other disadvantages to such an algorithm, such as:
Increased Complexity: The mathematical complexity of non-linear models can make them harder for users to understand and predict, which may lead to uncertainty about expected yields.
Higher Implementation Costs: Non-linear functions can be more demanding in terms of smart contract computations, potentially resulting in higher gas fees and a greater risk of coding errors.
Calibration Risks: If not carefully tuned, non-linear models might introduce abrupt changes in yield rates, which could inadvertently incentivize early unstaking or destabilize the reward system.
That said, several key advantages exist, so it's possible that governance will choose to upgrade the staking contract in the future. These advantages are:
Dynamic Reward Scaling: A non-linear model, such as a logistic S-curve, allows for steeper yield increases during key periods. This means that stakers could see more dramatic rewards as their stake matures, which can be very motivating during early growth phases.
Market Alignment: By adjusting parameters like steepness and midpoints, a non-linear model can more precisely align token emissions with market transaction growth. This can help maintain stable token value and better match incentives with real market conditions.
Enhanced Incentive Flexibility: Non-linear formulas provide flexibility to reward stakers differently based on their maturity, potentially creating tiers of rewards that incentivize long-term commitment more effectively than a linear model.
The primary goals of our staking program are twofold:
Encourage Long-Term Commitment: By increasing yields over time, the program incentivizes holders to keep their gD3L staked, ensuring they benefit from a growing share of the ecosystem’s rewards. This structure fosters a Nash Equilibrium where de-staking is consistently less attractive than remaining staked.
Stabilize uD3L Value: uD3L tokens are minted exclusively through staking, and their issuance is tied directly to the maturity of staked gD3L. The emission rate is designed to mirror market growth, thereby providing a stable medium of exchange within the Deep3 ecosystem. This controlled, transparent issuance helps maintain uD3L’s value even as the network scales.
Our staking parameters, including the yield rates and timing thresholds, are fully adjustable via governance proposals. This means that the community—through the D3L DAO—can fine-tune the staking model in response to market conditions and technological advancements, ensuring that the system remains both competitive and sustainable.
The initial staking contract (v1) utilizes the linear model, which has been optimized for simplicity and gas efficiency on the Ethereum blockchain. However, our platform is designed with flexibility in mind. When the community and our research indicate that a non-linear model would better serve our objectives, the DAO can vote to deploy a new staking contract. Importantly, the upgrade process is designed to preserve each staker’s maturity—ensuring that users do not lose the rewards they have accumulated, even as the underlying formula evolves.
This upgrade path represents our commitment to continuous improvement and community-driven innovation, ensuring that our staking program remains at the forefront of decentralized finance while adapting to new challenges and opportunities.
The Deep3 Labs staking program is more than just a mechanism for earning tokens—it is the engine that aligns incentives across our ecosystem. By rewarding long-term staking and tying uD3L issuance directly to market growth and staking maturity, we create a self-regulating, value-enhancing system. This program not only secures the stability and growth of our utility token but also empowers the community to shape the evolution of our entire platform through decentralized governance.
The Deep3 Labs governance token is tentatively named "gD3L".
The gD3L token is the cornerstone of the D3L DAO, granting holders fractional ownership rights and a direct say in the development and management of our AI ecosystem. With gD3L, you gain both economic benefits—such as sharing in the revenue streams and residual profits generated by the marketplace—and the power to influence key decisions, from approving model updates to setting economic policies. In effect, holding gD3L means participating in both the ownership and strategic oversight of the platform.
While a full breakdown of our planned token economics will be shared shortly, what we can confirm now is that the gD3L token will have a fixed supply, with no ability to mint additional tokens in the future. This fixed supply ensures that the token retains its value over time and provides a solid foundation for sustainable growth within our ecosystem. By design, the gD3L token underpins the economic and governance framework of Deep3 Labs, aligning the interests of all stakeholders and ensuring that every participant benefits from the platform’s success.
As a governance token, gD3L empowers holders with the right to propose and vote on important decisions affecting the D3L ecosystem. This includes determining marketplace changes, approving new model submissions, and even guiding economic actions such as treasury distributions. The transparent, blockchain-based governance process guarantees that every vote is recorded and that decision-making power is distributed equitably across the community.
In addition to its governance role, the gD3L token serves as a symbol of ownership, linking token holders directly to the platform’s performance. As the ecosystem grows and the marketplace sees increased transaction activity, the economic benefits for gD3L holders will expand correspondingly—reinforcing the token's value and the collective success of the D3L community.
The Deep3 Labs ecosystem is built on a dual-revenue model that harnesses both token price appreciation and traditional cash flows. This dual approach is designed to provide greater resiliency amid cryptocurrency market volatility while fueling innovation across our research and development efforts.
At the heart of our ecosystem are two ERC-20 tokens: the governance token (gD3L) and the utility token (uD3L). Both tokens derive value from their inherent utility and the robust demand generated by our platform’s use cases.
The primary drivers of token price appreciation include:
Perceived Value and Future Performance: As users and investors recognize the potential of our decentralized AI platform, speculation and market sentiment are expected to drive demand for both gD3L and uD3L.
Supply and Demand Dynamics: The fixed supply of gD3L and the controlled, variable supply of uD3L—governed by our staking program and DAO oversight—create a balanced system where scarcity and utility drive price discovery.
Broader Market Trends: Overall growth in the AI and blockchain markets will also contribute to token appreciation.
Additional planned use cases that support price appreciation include token-gated DAO participation, oracle node hosting, metered API and oracle access, and marketplace fees, royalties, and collateral requirements. These functions not only enhance the practical utility of our tokens but also ensure that as platform activity increases, token value will rise in tandem.
Relying solely on token price appreciation can be volatile. To mitigate this risk, Deep3 Labs also embraces a traditional, cash flow–based revenue model, leveraging both fiat and digital currency payments. This dual approach provides a stable income source that complements token-driven gains.
One key revenue stream will come from serving Web2.5 companies—organizations transitioning into the Web3 space—demanding advanced advertising and targeting capabilities. While these companies may prefer to pay in fiat rather than cryptocurrency, they are attracted to Deep3's sophisticated targeting tools that mirror the state-of-the-art methods used by leading Web2 platforms. By offering robust, cutting-edge marketing solutions that integrate seamlessly with Web3 technology, Deep3 Labs is positioned to capture significant revenue from this market segment, ensuring steady cash flow even amid token market volatility.
A core tenet of the D3L vision is to distribute the value created by our AI/ML pipelines more equitably among all contributors. Traditionally, businesses have reaped significant rewards from user data while offering users little more than free services. Deep3 Labs is committed to reversing this imbalance through our revenue sharing model.
Revenue sharing in the D3L ecosystem occurs on two levels:
For Platform Members: Anyone who holds gD3L tokens and participates in our DAO automatically becomes a platform member. These members earn rewards primarily through the staking program, where their staked gD3L tokens generate uD3L tokens. When these utility tokens are sold to customers for access to D3L assets, the proceeds are returned to the member, effectively sharing in the platform’s success.
For Data Scientists and Model Developers: Those who contribute to the creation and maintenance of our AI models enjoy additional rewards. They can set fee structures for their submissions on the D3L Model Marketplace, sharing in the revenue generated from model sales. This dual incentive ensures that both technical expertise and active participation in governance translate into tangible economic benefits.
Our revenue sharing strategy is designed to ensure that as the D3L ecosystem grows, the value generated by our AI models is distributed fairly among DAO participants, developers, and ultimately, the broader community of blockchain network users.
In summary, the Deep3 Labs monetization strategy is built on a robust framework that leverages both speculative token appreciation and traditional cash flows. This dual approach not only mitigates risks associated with cryptocurrency volatility but also reinforces the sustainable growth of our platform—driving continuous innovation, fair value distribution, and long-term success across the ecosystem.
The Deep3 community is the heart of our ecosystem. In the world of blockchain and AI, community engagement is not just a metric—it's the driving force behind innovation, transparency, and sustained growth.
In the crypto space, a vibrant community helps spread awareness, drives adoption, and supports a healthy secondary market for tokens. At Deep3, our community plays an essential role in advancing our mission. Feedback isn’t just valuable to us—it’s necessary. Your participation amplifies our impact, fuels discussions, and propels our ecosystem forward.
What sets Deep3 apart is our unique governance platform. Our community members don’t just use our tools—they help shape how AI is developed on Web3. You have a direct influence on critical decisions, from selecting the data used in our models to determining how they are deployed on dApps. Moreover, you control the economic backbone of our platform. Token holders have authority over key aspects of our token structure, including staking rates and circulating supply, ensuring that the ecosystem remains fair, dynamic, and responsive to real-world needs.
By being a part of the Deep3 community, you become an active participant in a transformative movement that is redefining AI on the blockchain. Your voice matters, and together, we can build a future where advanced technology is both accessible and equitable.
Below you will find links to all of Deep3's communities and social media.
Drop us an email any time at: hello@deep3.ai
