Ep3. The Tech Stack: A New Internet Architecture | The Post Web Audio Documentary

Speaker A: Foreign.

Speaker B: Welcome to Chapter three of the Post Web audio documentary In Chapter two we described how we believe we are transitioning to the Post Web and and offered a critical retrospective on a decade of Web3 its limitations, attempts at abstractions, but ultimately the promise it still holds. In Chapter three we will explore what exactly makes up the Post Web, its technological innovations and how they come together to form the Post Web Stack. An open source and highly modular stack of technologies based on the convergence of AI, IoT and DLT, creating a powerful flywheel of adoption. A flywheel that when it catalyzes, promises to pull in billions of users and trillions in assets and global GDP into a permissionless economic system and one that fuels a process of society wide hyper financialisation. So what is the Post Web stack? At a high level the Post Web is a new technological paradigm for the Internet based on the convergence of Uh3 foundational agents, distributed ledger technology as well as finally deepin decentralized physical infrastructure. Each of these have their own distinct technological and economic drivers of push and pull dynamics that complement one another's development and pathways to adoption. In summary, AI agents transform our interactions over the Internet by abstracting and managing tasks autonomously, driving demand for scalable infrastructure, uh and trustful verification mechanisms like dlt. Whilst DLT provides transparency, trust and accountability via immutable records and smart contracts and and AI agents and decentralized infrastructure enhance its scalability and resilience by distributing compute storage and connectivity for network optimization and transparent coordination. In aggregate they form a powerful flywheel that fundamentally reconfigures the Internet bottom up and infuses it with Web3's innovations of decentralization, digital property rights and user sovereignty wealth with ruthless efficiency to disintermediate and deliver better, faster, UH and ultimately cheaper web services. So you might be reasonably asking yourself, haven't Web3 Technologies been around for several years with no to little real impact on how the average person uses the Internet? Why is it so important now? As discussed in chapter one and two, the key difference is the Post Web merges a new agentic layer with an existing but improved Web3 stack optimized throughout by agents in the Post Web. The Web three stack of wallets, dapps, DLT and decentralized infrastructure become the foundation as its core primitives and economic rails, but must evolve to accommodate a very different new user. AI agents wallets will need enhanced programmability and data permissions to manage assets and identity for agents. Dapps require greater interoperability and adaptability to handle agent driven workflows and DLT must scale in efficiency, speed and programmability to support real time verification and trustless execution through a more native and efficient infrastructure to deliver more distributed compute, storage and connectivity. Whilst a token engineering that will coordinate economic activity like everything in the stack makes, needs to become more robust to handle the exponential demands of the agentic economy. So assuming everyone has a base understanding of Web3 technologies, let's start with what's new. In the Post Web everything begins with the user and their implicit or explicit intentions. Where the Internet increasingly organizes itself to fulfill intents optimally and contextually, from the appropriate user interface and, and degree of human involvement down to the network resources used to deliver it.

Speaker C: As Thomasin Genrola puts it.

Speaker D: So first of all there's understanding, okay, you know, what is the user's intention, what's the intent? So the way we do that today is by prompting, text prompting or voice prompting, right? So now you prompt this agent or assistant and it needs to understand the task you need to do and then find out, okay, what are the tools I need to actually solve this? So searching for the right tools to use, searching for the right content to curate or to collect to properly answer the question,

Speaker C: as Alex Pooch puts it,

Speaker E: that's when I think that the most important thing right now is to get the most accurate information in real time as possible. Because this is not about um, the UX anymore. I think like we've seen that like things that they're moving slowly, like far faster with the frames. But you don't need the fancy website, just a small application. You can interact with a, uh, tweet, with a chat, with something, that's enough. If we can integrate these small apps with AI, then we don't need websites.

Speaker B: All of this is negotiated through a user's sovereign agent within a secure and highly permissioned new privacy paradigm, providing the option for users to bring their own resources to complete tasks more efficiently and effectively. From data and AI models to compute and storage from their consumer hardware. For more personalized outcomes, the Post web will be delivered through hybrid architectures from locally hosted to edge and cloud infrastructure, giving the user more granular control. Now this might feel like a subtle nuance, but it is this that fundamentally flips the web on its head, transitioning us from the attention economy to the intention economy. From this core layer in the stack, the user layer, we're now able to directly delegate the more transactional aspects of the web to a universe of agents that can handle unparalleled, um, levels of complexity. Laser focused on delivering us the best outcome. But what exactly uh, are agents? Well, at a high level, agents, at least today, are autonomous, goal oriented, simple software programs that in aggregate form adaptive systems that can operate and evolve within complex and dynamic environments. However, in the Post web we go beyond today's already impressive agents to something much more powerful, something we're calling smart agents.

Speaker C: As Graham Barnes puts it,

Speaker F: what point are the agents going to start managing your interaction with real world services? And I think this is going to happen through Deepin, maybe something small at first. Hey, ah, we see uh, on your computer you have some available ram. Would you like to contribute that to um, uh, to something like Morpheus or just decentralized compute for AI agents? And you'll say yeah, and then it goes maybe a level up your wi fi, um, and then at what point does it start to become your electricity provider? Um, at what point does it start actually saying like, hey, I will pay all of the subscriptions, Like I've analyzed your bank account. Um, I tell you what, you're paying too much for Netflix. Um, I'll negotiate you a better deal.

Speaker B: Agents that possess the ability to interact with DLT and smart contracts directly, giving them greater levels of economic and even legal agency to represent all facets of our personhood. We'll delve deeper into the complexities and possibilities of agents a little later, but for now all you need to understand is the importance of having your very own agency agent. An agent to mediate with a universe of third party agents to translate your needs and intentions. Your sovereign agent is a personal assistant, confidant, helper and gatekeeper that will increasingly even be able to intuit and predict your needs, remove obstacles and help manage the complexity of the Post web through delegation, leaving you with a healthier or a more purposeful relationship with the Internet.

Speaker C: As Torbear puts it.

Speaker G: My understanding of Post Web is we're eliminating a lot of the user experience friction where we expect the user to be able to make sense of the nonsensical, where they're asked to do more than they really should. They just want to solve problems, they want to get things done, they want to connect to other people, they want to hit the entire hierarchy of needs. They want to be self actualized. And there's ways that Post Web can fix that. Letting AI work for them, making it easier for them to set direction for agents that live outside of themselves, that can do the things we don't think are human. The very rote activities that we're just, you know, we would rather be without so we can focus on more creative acts or more human acts, or more social acts.

Speaker B: What's important to stress here is that the consequence of delegation renders much of today's conventional web of search engines, apps and platforms redundant. Where you would otherwise spend hours navigating the attention economy, more often than not settling for convenient yet suboptimal outcomes, your sovereign agent will demand only the most relevant information, products and services communicated through minimally interruptive interfaces, and only when absolutely necessary for you to be involved, leaving you to enjoy the residual but powerful part of the Internet we're calling the thin web.

Speaker C: As Thomas Ingenrola puts it,

Speaker H: Post Web

Speaker D: is a very interesting concept. And the fact that we would have all this content be able to be generated on the fly means in some cases you may not even have websites. Agents could just take the data, um, and just generate the UI on the fly as needed for the user. Then you have even voice agents where um, the agents could just answer those questions. So we will definitely see a convergence of a lot of technologies together to be able to answer questions and help people solve things. But really we're entering this era where it's more of thought to action. So that's really where we're heading to.

Speaker B: We call it thin because relatively speaking, it will feel less burdensome and involved fewer websites and apps, less clicking, wiping and fewer tabs open. But we recognize its name could also be somewhat misleading because in fact what lies beneath is a significant set of new powerful innovations. A way of thinking about this elegant, yet technically complex new world is a shrinking of the web, but a simultaneous upgrade under the hood for supercharged Internet. Equally, we believe the thin web itself and, and our experience of it will lead to a reimagining of form factors and user interaction. Much like the smartphone redefined how we interacted with the Internet in the early 2000s. The post web will feel fundamentally and unmistakably different. We just don't know exactly how yet.

Speaker C: As Luis Oscar Ramirez puts it

Speaker I: today we have the framed media which is within the boundary of a smartphone or the boundary of a computer monitor. That boundary just exactly comes straight into your eyes and you get inside the content which is a spectrum like you see a completely entirely virtual world, but also it's just a, ah, completely synthetic reality that is mixed uh, with your daily life. In 20 years, in a way, the wife, the web dissolves in terms of the user experience and user interface. It will be embedded into our daily life and we don't even, maybe even notice the new generations that is the web. It's just a backbone infrastructure. It's there and everything is mixed.

Speaker B: What we do know is it will likely be made up of minimalist interfaces that will dynamically adapt to users intents, seamlessly blending information to tools and experiences while remaining accessible across today's existing devices as well of course as entirely new form factors and hardware designed natively for the post web era. In many ways this is an evolution that represents both continuity and transformation. And because of its more natural language and intuitive interfaces, it won't have the same generational lag of adoption seen in other leaps in information technologies like the smartphone or social media, with older users being left behind. However, we do believe, at least initially, the postware will be primarily experienced through a browser like interface, but one that has the ability to become dynamic, contextual, task specific and immersive when desired, with varying degrees of augmentation when appropriate. We think of this as a kind of smart browser allowing crypto to leap over the top of app um and game stores and other web two walled uh, gardens entirely while seamlessly integrating 3D, spatial and augmented reality capabilities and increasingly intelligent experiences tailored to the user's needs.

Speaker C: As Mike Charolambas puts it

Speaker J: and our thesis has always been that we live in a 3D world and finally the Internet is catching up. We are experiencing from early days this evolution of the browsers, this evolution in brands innovation appetite and how content is evolving to meet new standards, to create new kind of experiences, emotions and even provide new kind of metrics and KPIs for brands. And in the era where AI is now starting to really pick up and be part of our everyday and being able to make a lot of our process more efficient. We reached the point that we can now command the 3D to optimize command the 3D to take different kind of solutions and shapes and phases and then instruct it how to go and get deployed across the different uh, channels, across the different ERPs and CMS uh that we're working with.

Speaker B: So now let's deep dive into the post web second most important constituent agents in the agentic layer. It is through the agentic layer that AI agents can now autonomously fulfill user intentions, interact with APIs and Web3's decentralized applications and protocols to execute complex workflows with with minimal human involvement. Together these new layers and components enable seamless interaction, delegation and execution. They drive a transformative shift in how users and agents engage with the enhanced decentralized infrastructure and applications that form the foundation of today's Web3 ecosystem. In summary, the agentic layer enables smooth interaction between humans, machines and protocols. Now it's important here we pick up back on better defining agents to understand what those interactions might look like. Because many understandably conflate agents with Today's bots and Web3's solvers and crawlers. But we believe true AI agents are uh, autonomous goal oriented adaptive systems that can operate and evolve effectively within complex and dynamic environments. Whilst agents and bots can both execute tasks, their underlying design and capabilities differ.

Speaker C: As Lorenzo Cecilia puts it,

Speaker K: solverse is basically tailored around a uh, very specific task. Ah, agents is more towards the um, reactivity or the possibility to react, uh, to contexts that are changing. But the big difference will be that the agent is capable to keep the history of previous interactions, basically keeping the context given the um, access to new inputs initiate some type of reaction. And so this ping pong is what makes the agent different from a solver.

Speaker B: Bots are task oriented, built to perform specific, predefined and already understood actions. Agents in contrast are goal oriented, optimizing for overarching objectives that may span multiple interconnected tasks. Bots follow a deterministic decision making process where every step is predefined leading to predictable outcomes, whilst agents operate probabilistically where only the possible outcomes are known and the steps to achieve them are dynamically determined based on context, learning and environmental factors. Moreover, bots are inherently reactive, responding to inputs and external triggers in predetermined ways, while agents are adaptive systems capable of learning and adjusting over time. Whilst bots can simulate adaption through pre programmed behaviors, they lack the genuine autonomy or learning capabilities of agents.

Speaker C: As Dominic Williams puts it, if you

Speaker L: think of something like a chatbot, actually it's not quite as complicated as it might seem. You propose the contents of the chat so far and then the question is there are 50,000 words that could come next. Which of these 50,000 words is most likely to come next? And um, the chatbot will choose the most likely word, um, add it to the existing words and then sort of resubmit that as though it were the image. Okay, now given the original words plus this additional word, what's the most likely next word out of these 50,000 potential words? Right, so um, that's how it works. And um, if the chatbot always chose the most likely next word, um, you'd always get the same result. But in fact what they do is um, they shake it up a bit. So um, yes, sometimes it will choose the most likely word, sometimes the second most likely word, um, and so on. And of course um, the sequence of words that you add completely changes the output. But essentially the chatbots work very similarly to that image classification. The actual output is deterministically produced, but you know, it's using some kind of random seed and can draw the. You know, the Internet computer supply supplies deterministic randomism randomism to code, um, but it's using that random seed to decide which of those most probable next words to be selected. And that's why you get a different answer every time. And if you optimize inference using hardware like GPUs, you can also get some non determinism from the hardware. But there's nothing stopping AI inference being run in a fully deterministic way.

Speaker B: As hinted to earlier in the post web, we envision the emergence of a new even more powerful class of agents we refer to as smart agents. These smart agents will retain all the core characteristics of traditional agents, but with the enhanced capability to integrate interact directly with DLT and smart contracts, tapping into their functionality to execute actions on chain while also having their own actions transparently recorded through verification and specification. Smart agents are an order of magnitude more powerful and a core component of this layer in the post web stack because their AI systems uniquely capable of interacting interacting with DLT and smart contracts marked by greater levels of autonomy and new forms of agency and intelligence. So let's examine the foundations of smart agents, their deployment orchestration and their role in shaping the post web. To start with, on chain specification and verification and ah are two important functionalities that distributed ledger technology unlock for agents in the post web. On chain verification allows users to retroactively audit agents actions with transparency ensuring accountability and trust. And on chain specification enables users to define and enforce verifiable constraints or objectives for agents provably guiding their behavior in the post web. Together these capabilities are pivotal as millions of smart agents emerge economically transacting and executing tasks on behalf of users. By leveraging dlt, the specification and verification of agents ensure a secure and transparent framework for this new emerging autonomous economy.

Speaker C: As Leonard Doorlokter puts it,

Speaker M: the robotaxi comes and picks up the person at home because that robotaxi was ordered by the AI agent. And the robotaxi then being an uh, autonomous machine, but also having an AI agent probably that allows it to for example transact on the blockchain because it's kind of the function on top of the autonomous machine that is solely responsible for e.g. publishing when it's available for ride or um, you know, transacting with um, the customers or other AI agents that, that are booking it. And then you have the agents, you have the autonomous machines, you have the humans who all work together to create the most seamless experience, um, that, that you can imagine and the most efficient economic interaction as well.

Speaker C: As Emmanuel Aldegar puts it

Speaker N: very important to guarantee the security and the function of the network, meaning that what Web3 is bringing is the capabilities to have decentralized infrastructure who are working by their own, who are autonomous infrastructure. To allow this autonomy, you need the token to reward the contributor of the network. And I think this is very, very important because you have open source environments, uh, autonomous that no one can stop. This is guaranteed by the existence of the token which is rewarding the contributors to the network.

Speaker B: But what drives the intelligence of these agentic systems Today, AI encompasses various learning paradigms, each tailored to specific challenges and tasks such as supervised, unsupervised and reinforcement based learning. These paradigms enable AI systems to learn, adapt and perform complex tasks across diverse applications, be it centralized or decentralized. And they form the foundational methodologies for training AI models. In this context, agent typically follow a three step workflow of receive, analyze and act to interpret inputs, evaluate decisions and execute tasks, while continuously refining their approach through feedback mechanisms.

Speaker C: As Theo Priestley puts it,

Speaker O: when we're thinking about AI and agents in an organization, you know, the potential to have hundreds is there. We're not talking about a monolithic chatgpt sitting over in the entire organization. I mean you will have something like that, a uh, larger one sitting as a department function, but all the agents that sit underneath that will have been trained on very specific areas within that department or that function. And you could have hundreds. And that's, you know, that in itself is, well, one. Well, it might scare people, but they almost act like we do in terms of I'm speaking to you and you're telling me something and I'm going away and acting on it and stuff like that. It also means that there's a great upscaling opportunity again for the workforce. Rather than binning people. Again, you're acting as subject matter experts for individual agents and you're almost helping them become smarter by developing your own skills as an AI, uh, engineer in a sense, or a prompter or whatever you want to call yourself. I think that's key.

Speaker B: But to be truly effective agent based architecture, we believe we'll need to combine several intelligence components including perception, reasoning, learning and execution to facilitate autonomous, goal oriented and adaptable agents. As a consequence we believe smart agents in the post web can be classified across several dimensions. Their functional scope, form of ownership and method of orchestration. Functionally, agents can range from special purpose designed for specific tasks to general purpose agents capable of adapting to diverse domains and objectives. The ownership of agents in the Post web spans a spectrum of sovereignty from user owned agents that are fully controlled by individuals to platform agents that optimized for platform specific goals and fully autonomous agents who are self governing and optimized for independent incentives. And finally for orchestration, as Butean Lee

Speaker C: puts it,

Speaker A: a little bit more about the blockless technology, um, for being able to coordinate all those heterogeneous devices uh, across the network. We have a automatic orchestration mechanism that runs on chain. It essentially does the uber matchmaking between compute, ah, workload um, and the compute devices. Being able to contribute compute through the web browser, it's just making it a whole lot easier for uh, the users, for everyday users like for example my mom and my dad, to be able to be part of this shared economy.

Speaker B: Agents may operate at an individual level or all the way to collective groups of agents known as swarms.

Speaker C: As David Minosh puts it,

Speaker P: us humans are you know, general purpose to a large degree and then we kind of specialize in certain areas by choice and by happenstance. And that's one part I, I agree that um, we'll retain specialization and that we have these sort of emergent outcome outcomes which are the result of agent interactions. So oftentimes the user might um, consume something which isn't produced by a single agent but by an entire swarm of agents, a multi agent system.

Speaker B: In summary, while smart agents will vary wildly in capabilities and competency, they are generally goal oriented as in they are capable of strategic planning, clarity of purpose and um, precise execution. They are autonomous, capable of contextual and situational awareness and have powerful economic agency, but are also adaptive with high degrees of sociability feedback and have adaptive memory

Speaker C: as Humayunshaikh puts it

Speaker Q: all about. But it's now becoming very clear that the whole Internet, the whole web is going to change and the way it's going to change is looking more likely to be in the agentic direction.

Speaker B: However, it's important to remember so far we've largely talked about agents and their capabilities at an individual level. But what differentiates agentic AI from other forms of more monolithic AI is that they aren't owned and controlled by single corporations and therefore not top down hierarchical systems, but kind of free range bottom up and open systems where agents compete and collaborate across niches, but in aggregate can give the sense of something close to an AGI like intelligence through the UX of your sovereign agent.

Speaker C: As Emmanuel Eldeger puts it, we believe

Speaker N: in Axon that collaborative AI will outperform siloed AI because companies are suboptimal coordination mechanism in the digital world. So imagine tomorrow's technology instead of a company that gives thousands, maybe millions of people the power to share their data, their models and their infrastructure that would democratize AI at scale and represent a huge value creation, right?

Speaker B: However and importantly where the intelligence is in fact not concentrated in a single AI, but instead a uh, diverse universe of actors. Some of these agents are more responsive and assist users by gathering information and waiting for further direction. Others more operative in that they anticipate users needs and take the initiative within defined boundaries. And others with high degrees of autonomy will manage tasks and make decisions with no to little human input. In summary, smart agents will operate across a range of social behaviors to both compete and collaborate. And in many cases agent to agent to negotiate outcomes, optimize resources and achieve goals. They would engage directly agent to protocol to execute transactions, manage smart contracts and interact with blockchain infrastructure. And they may even interact with or spin up their own daos. Agent to DAO to pool resources, coordinate activities, find collective alignment and um, manage on chain governance or interact with external devices and machines. Agent to device to leverage resources and enhance their capabilities to affect the physical world. The potential for agents when working with DLT allows for a civilization scale upgrade to how we organize and coordinate economic activity. Many times transformative than the industrial revolution was to society. And, and we'll see entirely new emergent behaviors and properties that are difficult to predict. But it's fair to say that decades of intention theory and designing and playing massively multiplayer online games with agent based systems, as well as a healthy dose of anthropomorphizing, can give us a sense of the direction of travel.

Speaker C: As Aaron MacDonald puts it, the short

Speaker R: term looks like a funner version of games where there is more dynamism, where the game quote unquote evolves organically, experiences become hyper personalized, then the next stage of that becomes more immersive. And this is kind of like if you look at what recently came out, we're one step away from him, probably not needing the screen. And so now you're connected to this digital, immersive, generative, hyper personalized world, your social interactions becoming immersed in those and then you're only like a hair away from the matrix at that point.

Speaker B: In the post Web wallets transcend the relatively simple tools for storing digital assets, cryptographic keys or just facilitating transactions and enabling interactions with dapps, but become the control center for personhood and resource management for a dramatically expanded itinerary of assets and capabilities. You guessed it, we're calling them a smarter uh, wallet as Jasper Demeyer puts it.

Speaker S: So smart wallets are far more advanced than today's cryptocurrency wallets. They not only hold the cryptocurrency we store today but they will virtually become the gateway for digital self identity and outsourcing economic activity. In the post swap these wallets will not only hold again the cryptocurrency assets but obviously also RWA which you already see to some extent. But the novelty really is the personage which is introduced within the wallet. With personhoods we believe that's personal data part uh, of your identity um, but also part of your data. And all of this is stored under an umbrella of consent and management. So consent and management will be very important because it dictates to what extent agents can use specific assets, use your data, uh, um, and what type of activities they uh, can ultimately execute. So, so you can look at wallets in the post web as this control center through which human users ultimately decide to what extent economic and social activity can be automated by agents.

Speaker B: Bolstered by privacy enhancing technologies and increased levels of functional forms of user agency and representation, wallets will serve uh, as the first foundation for interaction across the post web stack. They will carry context and store memory and prompts enabling trustful personhood and personalization for agents to securely access necessary permissions and autonomously execute tasks on the user's behalf. Whilst there are technically already smart wallets today, we have significant concerns about their reliance on centralized infrastructure where most still depend on cloud hosted relayer networks and off chain systems for critical functions like transaction processing, recovery mechanisms and notifications, creating single points of failure and exposing wallets to risks such as censorship outages, security and privacy vulnerabilities. And whilst that's already problematic today for users, it's even more so when delegating tasks with economic value to end agents with increasing levels of autonomy and complexity. So it is our belief the Post web requires a significantly more mature and decentralized architecture to ensure secure trustful systems we can feel comfortable delegating greater levels of our sovereignty and personhood to. However, whilst abstraction is critical for adoption, the interface for human users is still critical as we will continue to need to access wallets to directly manage some aspects of their personhood and will likely always need to provide a degree of human in the loop consent for high value or regulated activities and to continuously teach character and social function to our sovereign agents to better represent our unique personalities and preferences.

Speaker C: As Thomas Ingenrola puts it,

Speaker D: in a web3 environment, um, your agents would also have their own digital uh, identities that they control which the owners of those agents should be able to uh, manage in case of loss, um, of keys so that there's account recovery. But of course these agents should be able to send transactions on chain or initiate transactions. One key thing there that we are also ensuring we build is what is known as human in the loop. So for certain things, when agents are about to perform a very complex task, um, that needs a high level of oversight, the agents can actually call out back to another agent or to ideally a human to actually approve or decline um, an activity or an action before it actually happens. So you can actually specify a threshold. If the amount you're about to send is above 100,000, ensure you get this verification, this sign off from the following users or if you're about to send an email to you know my list of investors, right ensure that I get uh, one of the co founders to actually review you know, that email.

Speaker B: Taking this into account, we think it's likely a hybrid approach with sensible degrees of centralization that, that will be utilized by Web3's Ah, early adopters and innovators to improve onboarding and overall usability whilst gradually removing the risks of delegation in a centralized paradigm. This transition will be further accelerated by advances in privacy enhancing technologies otherwise known as pets combined with depin enabling highly functional wallets with minimal reliance on centralized systems. This transition will safeguard the growing assets and interactions within wallets ensuring resilience and alignment for more trustful systems as Torbear puts it.

Speaker G: Will they be able to route around user consent to be able to access these systems? Building on top of a confidential compute foundation means you have better control as a user and as a developer being responsible for the safety security of those users. That's a level of control that you want to have. If we build on this public by default foundation for this AI convergence with Web3 and beyond, it's already building, not just building on sand, it's building a dystopia. So what we're hoping is that people start to take the privacy risk seriously in AI first maybe because it's such uh, a tangible use case today, but mostly because if we don't take it seriously now by the time we do, it'll definitely be too late.

Speaker B: However, even with improved decentralized systems, some wallet functions may forever remain partially centralized due to unique performance requirements such as ultra low latency or high throughput applications. It is our belief that while the majority of infrastructure should decentralize to safeguard critical assets and interactions, retaining centralized components for specific use cases could be an acceptable trade off for for enhanced usability and adoption. So far we've alluded to personhood in general terms, but it is in and of itself an important and distinct technical layer in the stack that encapsulates identity, reputation, personal data and privacy, and that forms the user's unique personhood to allow allow them through their wallets to dynamically balance control, consent and secure interactions. Perhaps a more succinct way of expressing personhood is the comprehensive representation and management of a user's identity, reputation, data and privacy. This core part of the stack allows users to control their digital footprint making manage consent and permissions, and engage securely and privately with AI agents and decentralized services. But in order for personhood to have a meaningful impact, it needs to operate within a comprehensive and somewhat universally accepted framework to define identity and data privacy in a legal sense across, uh, geographies and jurisdictions.

Speaker C: As Al Morris puts it,

Speaker T: and that's actually a really important factor for privacy right now. When you use OpenAI or something, you're actually giving all of your private information to OpenAI, the company who's then looking it through with their model and then deciding what to do with it. Um, so the way that you fix privacy is by taking these much smaller models and running them locally on people's devices with access to lots of contextual information.

Speaker B: Now, a critical aspect of personhood is of course data and data sovereignty. One of the Original promises of Web3 the fact today personal data is still stored in silos on Web2 platforms in the cloud, even by Web3's supposed Dapps shows it's not an easy thing to solve at scale. Not only is this outside of the user's control, but much of the information is duplicated and outdated, undermining not just their sovereignty but more practically application effectiveness in the post web personal context. And therefore this data is the final mile for effective personalization for better and more optimal user outcomes where personal data will be treated as a bearer asset, securely stored within wallets. As more economic activity moves on, chain wallets will increasingly house transactional and behavioral data, uh, evolving from basic information to complex user profiles. This enriched data, uh, will enable highly personalized services while being safeguarded through robust privacy measures and granular automated consent mechanisms managed by consent and privacy layers. The Post Web will enable agents to use dynamic and um, portable data across dapps enriched by contextual information information to optimize information distribution in real time based on changing user preferences, locations and circumstances. At the same time, agents will preserve necessary levels of privacy by unlocking private verifiability through the use of privacy enhancing technologies. Building on personal data, identity and reputation can be synthesized into actionable metrics to form a a bottom up foundational system of trust in the Post web where identity provides a more static form of verification acting as a reliable anchor and reputation serves as a dynamic, cumulative and real time measure of credibility evolving out of users ongoing actions and interactions in the Post Web over time. However, it must be highlighted that the continuous interplay between static identity and dynamic reputation is introduces its own challenges in ensuring both can coexist seamlessly, especially in decentralized systems which aim to balance verifiability with more regulated kinds of activities with degrees of privacy and pseudonymity. Identity and reputation in the Post Web will extend beyond individual users becoming contextualized within their social graphs as UH social systems increasingly move onto DLT and through trends like socialfi. By incorporating relationships and interactions within the user's network, these systems can create more nuanced and context specific peer to peer forms of trust. The importance of identity and reputation as the enablers for trust based economic transactions in the Post Web cannot be overstated.

Speaker C: As Charlie Varley puts it,

Speaker R: what is

Speaker H: the route to the uh, mass adoption of Web3? Typically the general perspective in the space is that it will be through things like stablecoin payments and uh, decentralized identity, things that you almost never notice. Um, and clearly crypto and web 3 and blockchain provide a better vehicle for those types of use cases. The reality uh, is that while those will lead to the quiet adoption of crypto, if we want people to graduate from normies into degens and essentially take this next chapter in the decentralization of finance, that's going to be through more web3 native concepts like defi, uh, like social fi.

Speaker B: So now we've covered off agents, personhood, um, and wallets, let's move on to decentralized applications otherwise known as dapps, which really give purpose to the Post web stack by facilitating the coordination of its social and economic activity. In Contrast to today's Web3, the post Web uh divides its universe of decentralized applications into two distinct layers, the functionality layer and a peer to peer transactional.

Speaker C: Layer.

Speaker B: Whilst both of these layers are interconnected, they are distinct in purpose in the Post web. The peer to peer transaction layer is an evolved set of defi protocols like DEXS that orchestrate the underlying exchange of a growing range of assets and the functionality layer of DAPPS that coordinates goods and services on top using DLT smart contracts to ultimately disintermediate inefficiencies for both agent to human and agent to agent outcomes. This specialization means dapps will either one coordinate the exchange of goods and services at the functionality layer or two orchestrate the direct exchange of assets in the peer to peer transaction layer. As a result, not every DAPP will always need to build its own pen P2P transaction layer fostering modularity and scalability as more and more economic and social activity, whether real world or currently intangible value come on chain and become tokenized assets. Through a process we call hyper financialization. Peer to peer transaction layers become increasingly core to the functionality of applications and the post web as a whole, eating away at today's centralized apps and mobile marketplaces. Furthermore, the peer to peer transaction layer is pivotal in reshaping how transactions are orchestrated, enabling more granular and dynamic models such as pay as you go and pay with anything that is compute or data. Uh, moving beyond traditional subscription based systems. As Jason Potts puts it, so we're

Speaker U: entering into a digital economy and the fundamental thing about a digital economy is everything in it is computable. All of the interactions are fundamentally computable, which means that anything that has compute capabilities can interact with anything else that has compute capabilities and behave as if they were doing economic transactions. For thousands of years that has been limited to human to human interactions. We're just at the cusp of being able to massively expand the number of economic agents because of this enormous explosion of compute capability into modern economies. That's what's coming the computable economy.

Speaker B: As discussed earlier, whether or not humans directly interact with dapps will depend on their needs. According to Maslow's hierarchy, higher needs will still see humans want to interact with dapps at the functionality layer, uh, through the thin web, while the lower more transactional needs will be delegated to agents that take care of it autonomously to access the core service directly through a combination of other agents, smart contracts and APIs. The difference to today's Dapps compared to the thin web is that they will be supported and generated by AI on the fly in a hyper contextual way through a combination of image, text and video as well as co generation to create a tailored interface for the user based on mode environment, be it social or individual and preferences whether it be maximalist or minimalist interfaces.

Speaker C: As Ilya Polisukin puts it

Speaker K: m now not websites right? You'll have not just APIs, you'll have intelligent agents that are able to satisfy intent or provide context for your model to provide the right information. The front end itself is going to be generated on the fly on your phone. You don't need anyone to build anymore the apps. There's no apps anymore.

Speaker B: As discussed previously, this will be streamed through a browser like format that is highly intuitive and portable, most likely reducing the dependency on the local hardware requirements of users devices through edge services. Unless of course the user wishes to push the limits of experience requiring more innovative costly specialized hardware for greater levels of inference immersion and intelligence. Something we'll talk about when we get to the resources and infrastructure layer we believe in most cases it will likely be voice driven rather than digital interfaces using digits fingers to interact, which requires the creation of semantic maps between agents and humans alongside a whole host of other technologies we'll explore in later episodes and over time will evolve into brain computer interfaces where thoughts and their respective neural patterns both drive interactions and help visualize the interface in the mind's eye.

Speaker C: As Trent McConaughey puts it,

Speaker V: for humans to reach human superintelligence, there's two paths. There's uploads and there's brain computer interfaces where it's sort of like we accelerate bci. The uploads are a bit farther away. So it really comes down to accelerating bci. Acknowledging if we want to accelerate bci, we're going to have to have these high bandwidth interfaces of ourselves to these silicon devices and we are going to want to interact with the cloud, but at the same time we want to make sure that people will be very very scared if their thoughts leave locally and instead hit the servers of Microsoft or nsa. We need to actually preserve Cognitive Liberty. Our thoughts need to um, stay with our keys. Not your keys, not your thoughts. This is where Web3 can play a big role. Um, in fact it might be the killer app for the future. Cognitive Liberty as the killer app for Web3AI, decentralized AI, all of that.

Speaker B: Where usability and intuitive interface were previously a core focus through abstraction and orchestration layers, the differentiator between startups now will mainly be driven by their core functionality, removal of third party value extraction and network effects, making a broader range of dapps more economically viable. As alluded to earlier, the phenomena of Hyper financialization sees assets not traditionally viewed as financial, especially intangible assets, being quantified in monetary terms and leveraged for maximum financial gain. The push towards monetizing everyday assets will further drive the adoption of blockchain technology. And through tokenization, these assets become more financially productive, offering attractive investment returns. Those equipped with the financial literacy in this new paradigm and the ability to execute on it will not only just be able to broadly participate in the web as an economic cycle, but also compound their wealth and actively become shareholders in the tokenization of the post web stack as a whole. It is our hope that through the post webs more intuitive and inclusive natural language LLMs and agents, we see a leveling of the playing field by enabling basic financial literacy and participation for everyone at scale. These tools empower individuals to manage and leverage their tangible and intangible assets and wealth, ensuring broader access to economic opportunities and minimizing huge disparities in wealth and knowledge.

Speaker C: As Graham Barnes puts it,

Speaker F: I've tried the trade in the past, but I just can't. There's a certain mentality I think that's required to do it and to do it well. What I realized is that these institutional traders, these market makers, they're not doing judgment calls. I mean they are savvy professionals who are referencing like the latest literature around financial management to make decisions and really, really good at quantifying things like risk and potential upside and using that to make decisions around their portfolio. Now if you want to have access to that kind of intelligence, you need to have a family office or you need to run like your own market making firm. And there are people that have that luxury but they are very, very wealthy individuals. And my excitement around this defi use case and using agents to manage essentially manage your financial um, activity on blockchain is that that can be democratized.

Speaker B: So we're now at the distributed ledger technology layer. DLT, the technological innovation underpinning Web3 is a deep technology domain, so we won't go into it at a granular level here. Instead we'll rather focus on the potential they and tokens play as the trust layer economic rails coordinating agentic economic activity in the post Web. However, at a high level, as the post web transitions to a world dominated by autonomous agents, machine to machine interactions through tokenized economies, DLT provides the deterministic backbone necessary to ensure transparency, verifiability and accountability across systems. As of today, DLT is used by more rudimentary agents for ownership, distribution and tracing of transactions and state changes. However, looking ahead to billions of smart agents. They will all have wildly different needs and specifications for the DLT they carry out their work over. So the DLT stack itself will have to become more modular to optimize for specific functions, and so require seamless interoperability between its various layers. But ultimately creating a more resilient infrastructure where components can evolve independently. What's important to stress is increasingly agents themselves will decide which DLT they will want to transact on, begging the question, what are the different parameters they will choose from and what might become some of the generalised norms?

Speaker C: As Ben Goertzel puts it, they can

Speaker W: decide to adopt a certain consensus mechanism for a certain purpose and another one for a different purpose in the same or an overlapping shard. So if you're doing like swift code stuff, uh, when you're sending huge amounts of money around around, you pretty much want a ledger, right? And it's okay if it's slow because you want double trickle quadruple checking on on the million dollars you're sending here or there, right? Whereas if you have a population of AI agents sending ideas around, or if you're sending micropayments of a fraction of a US cent around for a decentralized social network or something, you're okay if some frauds are only identified after the fact actually, right? Which after all is uh, how MasterCard and Visa work too, then you can have a much faster and looser consensus mechanism.

Speaker B: Whilst DLT offers transparency and trustless environments for agents, they also have inherent limiting factors that influence when and how agents will choose to interact with them at all, from latency to cost and uh, other technical limitations. For example, DLT often requires confirmation times to validate transactions, introducing delays that may conflict with smart agents real time decision making needs on chain interactions can become overly costly due to gas fees and transaction costs. In particular for smart agents handling low value high volume tasks where the cost may outweigh the benefits of trustless execution. The post web request requires agents to interact across multiple infrastructure layers such as oracles and cross chain protocols. Managing this complexity for certain tasks may be beyond the capabilities of simpler, more narrow agents, limiting their ability to ever operate on chain themselves. And uh, technological limitations such as blockchain scalability can constrain the ability of smart agents to interact with DLT for high frequency transactions or complex multi agent interactions. We believe the decision for agents to interact with DLT ultimately hinges on the value of the action versus the cost of executing it on chain along with the requirement for consensus. It's unrealistic for us to assume that every single state change and interaction, whether human agent or agent to agent, will happen on chain. The decision to make the AI agent interact with DLT or use alternatives such as optimistic assumptions depends on a variety of factors that balance practicality, costs and functionality. These vectors are vital considerations that define the frequency with which agents engage with DLT to optimize for efficiency and security. Remember in the post web, many of these vectors and the ultimate decision to interact with DLT for a specific transaction will be automated and uh, mainly at the discretion of the agents themselves. However, in order to do so, agents need rich contextual information to make these decisions autonomously. Understanding these vectors is essential for system design and deciding whether specific use cases across industries such as such as finance and healthcare are uh or aren't feasible. Given the existing limitations of AI and dlt both today and in the future, it's important to strike the right balance between on chain assurances and off chain flexibility. As a consequence, we expect to see more experimentation and economic flow to variations of classic blockchains such as dags, decentralized acyclical graphs and other new consensus mechanisms which, with even quantum computing adding to the complexity. And we also need to allow for agents to operate on a spectrum of trust ranging from optimism based solutions to something called zero trust paradigms. Optimism is what's used today by layer 2s for ex optimistic rollups and is a security approach that assumes most interactions are legitimate by default, verifying only when disputes or anomalies occur. While zero trust is a security framework that operates under the principle of never trust, always verify, requiring continuous verification for every interaction and access attempt. Zero trust, otherwise known as zt, provides the highest level of security through continuous verification, treating every interaction as potentially untrustworthy. Um, but obviously applying ZT universally to AI agents is impractical due to the significant resource demands and technologies limitations.

Speaker C: As Mahmoud Shihata puts it,

Speaker Q: everybody's saying zqb this, this and that zqb it's a thousand x overhead compute to prove something like a model today. So we're using kind of like um, a optimistic uh, approach using proof of inference. The idea actually comes from the famous arbitrarium optimism chains. And the idea here is that you could assume something is true unless someone claims otherwise.

Speaker B: So we believe it's about striking the right balance between efficiency and security. But we think given its costs and performance benefits, it is likely that optimism will play a significant role as a security assumption for the post web for the foreseeable future. In fact, we Anticipate a new consensus formation tailored specifically for agentic activity, where interactions between AI agents are validated through mechanisms called agentic roll ups. Designed to balance the efficiency of optimism with the precision of targeted verification. This approach, we believe, will enable scalable and secure coordination amongst agencies. Moving on to the role of tokens in the post web, it's important to stress the difference in tokens in the context of classical AI and DLT in the post web. In AI, tokens refer to units of text that an AI can process, whilst with DLT we're talking about units of value recorded on a blockchain. But more than that, tokens are an economic and governance primitive that drive alignment and enable decentralization and seamless interactions. In the post web, tokens play a pivotal role in enabling decentralized systems to function effectively. Acting as economic and governance primitives of an agent driven ecosystem, tokens are the backbone for aligning incentives, distributing ownership and facilitating seamless interactions across increasingly complex networks. Unlike today, tokens will no longer be missed merely speculative assets, but critical commodities based on real world supply and demand. And the fundamental mechanisms for orchestrating value, governance and agency in the machine to machine economy.

Speaker C: As Demetrios Chatian Agnosto puts it.

Speaker X: And what happens when financial uh institutions can start making fundamental analysis is that from there you will be able to. You start seeing things being maybe price similar to like how an Nvidia stock being analyzed and price as well. But from there it opens uh up a new world for like financial products being developed and launched in terms of like certain indices, proxies baskets of new funds that the real world can start investing in.

Speaker B: But importantly we see a handful of key roles for tokens in the post web. The most obvious being as a medium of exchange. Tokens enable efficient transactions in a machine driven economy bypassing traditional financial bottlenecks. Enabling seamless financial transactions is crucial for agents to operate autonomously and efficiently. In the post web. The medium of exchange agents use will define their ability to execute tasks, align incentives and interact within ecosyste.

Speaker C: As Emmanuel Aldegar puts it,

Speaker N: I think it's very very important to guarantee the security and the function of the network. Meaning that uh, uh, what web3 is bringing is the capabilities to have decentralized infrastructure who are working by their own, who are autonomous infrastructure. To allow this autonomy you need the token to reward the contributor of the network. And I think this is very very important because you have open source environments uh, autonomous that no one can stop. And this is guaranteed by the existence of the Token which is rewarding the contributors to the network.

Speaker B: The programmable nature of tokens makes it possible to create various types of mediums of exchange for agents to operate across the stack, each offering different degrees of freedom or restriction. While token programmability will eventually create a full spectrum of mediums of exchange from general purpose, purpose built and a hybrid including things like CBDCs, that is central bank digital currencies. Again, all these will increasingly be selected by agents based on tailorability, interoperability and liquidity. Linked to programmability. Tokens provide a cryptoeconomic framework to create game theory for agent behavior, using primitives like staking to ensure commitment and slashing to penalize incorrect or harmful actions. In the post Web, staking and slashing mechanisms are fundamental for ensuring the reliability and accountability of agents. When agents are required to uphold a certain standard of performance, the hosts or the agent itself can be asked to stake tokens as collateral, creating a financial incentive for accurate and honest behavior. If an agent produces incorrect outputs or acts maliciously, its stake can be slashed, enforcing economic penalties for misconduct and uh, maintaining trust within the system.

Speaker C: As Kevin Lynn puts it,

Speaker Y: when I mention programmable, it's like we can design game theory and uh, mechanism design and also the ability, or we can also observe what things are happening and we can track all the transaction so we can actually see the precise result of the parameter change or the mechanism change. So I think that's what I'm so excited about DeFi and Web3 and all the smart contract programmable economics as well.

Speaker B: These mechanisms are further strengthened by systems designed to make agent outputs verifiable and accountable. Each output is digitally signed and recorded on chain, creating an immutable and auditable trail of activity. We're still in the early days, but we soon expect to see projects specializing in agent incentive design as they handle increasingly important types of assets. Another way tokens are beneficial in the Post web is around network effects. Until now, networks on the web have primarily benefited platforms boosting service quality and user experience. But this has come at the cost of competition and this lack of competition often results in high user switching costs and platform driven payoffs. In the post Web we expect the core advantages of network effects, enhanced service value and connectivity to remain. However, due to modular services, interoperable AI driven agents, decentralized infrastructure and user ownership of data, the restrictive internetwork effects will diminish. This shift preserves the benefits of network effects while removing the negative externalities, fostering a more user centric and open system whereby tokens incentivize participation and create compounding value as networks grow. And finally, with tokens it comes to governance. Decentralized governance over economically important infrastructure and protocols has been one of the core promises of web3. Tokens have played a central role in this vision, promising to act as the primary mechanism for voting and incentivizing participation and the distribution of power among stakeholders. And uh, whilst this is true in practicality a decade on, there are still structural issues with tokenized governance, including low participation, centralization of token ownership and inefficiencies in decision making resulting from the often intuited cost benefit analysis of users in participating. But these early challenges will be removed by agents themselves. Driven by automation of the governance process through delegation on chain AI agents will streamline governance, enhance scalability and empower both human users and agents in ways previously unattainable. Like many things in the Post Web, it's important to imagine users are often represented by agents with delegated roles and responsibilities based on specialist capabilities, similar to how representative democracies function today. And it's no small thing, which is why we have a specific governance layer to the stack. The Post web is actually an upgrade to humanity's governance capabilities to move beyond industrial era frameworks of orchestration like the corporation and analog cooperatives or even the nation state, towards more dynamic technologically enabled governance underpinned by collective action, as Theo Priestley puts it.

Speaker O: So I've been digging into this idea of network states. It's been proposed by a number of bench capitalists that using blockchain and crypto and decentralization, you can essentially build like, almost like micronations that are decentralized and run their own schooling systems, have their own governance built in daos and things like that. It's very much an accelerationalist attitude towards this. Now I looked at that and I thought, well, instead of having network states, why don't we just become stateless? And it's almost going on this principle that well, there's only one planet and all borders are basically artificial. But how can we use that as a stepping stone towards we're looking at post web? How do we go post economy in a sense and look at well, what do we need to get to a post economy, post scarcity Type future?

Speaker B: While DLT during Web3 gave users the right to participate, it fell short of enabling effective execution. Governance in the Post Web resolves this by introducing automated mechanisms powered by AI agents and decentralized infrastructure, scaling participation and orchestrating activity at unprecedented levels, as Nick Allman puts it.

Speaker Z: So DAOS have a huge kind of coordination bottleneck and most of that's informational. So I see LLMs as stepping into this space to how many people understand each other. So you can kind of funnel decisions into a uh, high legitimacy, uh, outcome. And in terms of where it's emerging at the moment, I think we're at this kind of infrastructural phase. So I feel like we're just starting to emerge out of that in Web three where it's starting to move more to the social layer. And I actually think AI is going to be the huge unlock for that. And then once we're at the sort of agentic layer, they can then communicate to each other in a kind of network of agents. Um, so I'm calling this a network of experts where we have these individual agents and they kind of, they're brilliant at matchmaking. So it's like I want this information, then they can go and run around a network and find the information and bring it back. A big part of the job in my mind is um, finding these informational points that we move to and agents move to and finding this kind of like third space of interaction.

Speaker B: For centuries, governance oscillated between free market driven individualism and its counter forces in collectivism and regulation, often swinging to extremes due to the limitations of analog systems. We believe the Post Web finally resolves this tension by embedding governance directly into decentralized uh systems and constantly balancing these opposing forces into optimal states. Just as economic activity and mediums of exchange have been ported into the digital realm, the Post Web finally revolutionizes and integrates governance directly into the Internet. Tokens empower decentralized decision making, ensuring equitable control and stakeholder alignment across the full post web stack, creating better outcomes, removing principal agent conflicts and providing the long tail of users rather than a small group of corporations with a credibly neutral Internet. Linked to governance is the ownership of powerful agents and AI foundational models themselves as well as the data that powers them. Rapid advances in AI are set to disrupt labor markets significantly.

Speaker C: As Nick Allman puts it,

Speaker Z: if we actually just figure out the mechanisms for collaborating at scale, we could unlock a world of problem solving that we could never do in like centralized structures like the academy or on social, um, media networks, for example. I can imagine there being a hybrid structure at play here, but I do think the classic equity structure creates basically a civilizational level principal agent problem.

Speaker B: While AI can enhance productivity for some workers, it will replace tasks performed by others and transform nearly all occupations to varying degrees. Extending far beyond traditional software, AI's impact spans routine tasks to complex decision making, reshaping industries and fundamentally changing the nature of work. At the same time, AI will become the largest deflationary force since the Industrial Revolution. By automating tasks, reducing costs and driving efficiency, its benefits will eventually reach consumers through lower prices. However, it may take years before this value creation is fully passed down, as much of it will initially be extracted by those who own the AI models that drive the decisions. This is why the swift distribution of ownership of AI models themselves is essential to democratize, access and share the value generated more equitably, rather than concentrating it among a few large platforms and its shareholders. Infrastructure and resources are the backbone of the postweb. Encompassing data, compute, storage, connectivity, latency and energy. They collectively enable scalable, efficient and resilient system operations by leveraging decentralized physical infrastructure networks, otherwise known as depin and tokenized systems. The post web orchestrates resource allocation dynamically transforming idle capacity into actionable utility while, uh, lowering barriers to entry for hosting infrastructure. These components are becoming increasingly granular, with both demand and supply size driving innovation. Historically, the supply side of cloud infrastructure was dominated by centralized players with access to deep capital pools essential for hosting in Web3 and by extension, the post Web. This paradigm shifts in two fundamental ways as barriers of entry to host infrastructure and resources go down drastically. One is capital formation, where tokens enable lowering barriers to entry and allowing community communities to fund and build infrastructure collaboratively. And the second, as described earlier, is orchestration, whereby through tokenization incentive mechanisms, game theory facilitates the coordination of individuals to host and maintain infrastructure efficiently. On the demand side, the use of tokens addresses the cold start problem faced by early stage infrastructure and resource networks, which almost always rely on network effects by incentivizing early adoption through subsidized token distribution.

Speaker C: As Trent McConaughey puts it,

Speaker V: what's the largest computer in the world? Answer, the largest computer in the world by far is Bitcoin. How some dude or team of people wrote an algorithm into software, published that software onto a random mailing list, and from that mailing list enough people started running this and believing that hey, maybe there's value in this thing called bitcoin which incentivize them to keep running it right? And bit by bit it snowballed exponentially. Within a few years there was already hundreds, thousands of people running the nodes. Now there's tens of thousands, hundreds of thousands, and there's this sort of collective belief in the value of Bitcoin itself. These days whatever, 50k, 60k thousand US dollars that it's worth going for, right? And that's what makes it the largest network in the world was there's enough people running compute because they believe that Bitcoin is valuable. And to me what's cool about this is that it really highlights the power of incentives. Right? And then what's the heart of this? Uh, it's one of the superpowers, or if not the superpower of blockchain, which is you can get people to do stuff by giving them tokens, right? And that's a wildly powerful idea. So you give them tokens to run the Bitcoin node. You can give them tokens to train an AI model, right? Or you can give them tokens to run inference on an AI model.

Speaker B: These financial incentives attract new users to networks that are not yet large enough to sustain strong network effects, helping to bootstrap network growth and strengthen the product in its early stages. This decentralized approach to hosting infrastructure and resources builds on the principle of Web3, fostering a more resilient, scalable, user centric and driven infrastructure for the post Web era. However, the adoption of Deepin in the post web requires careful balancing of UH3 critical cost, efficiency and risk, each playing a pivotal role in determining its viability over traditional infrastructure. While Deepin reduces reliance on centralized capital, higher initial or operational costs can impact adoption where uh performance and scalability may lag behind traditional systems requiring optimization to remain competitive. Deepin mitigates risk from single points of failure, but the tradeoff must justify potential costs or efficiency losses. These factors ultimately determine the practicality of a Deepin implementation and will always require careful balancing. While we do not advocate for everything to be hosted in decentralized way, and as economic activity increasingly flows into the post web, including sensitive information and personhood, we believe their risks should become a more meaningful consideration than they are today, driving the adoption of Deepin as a solution in the post web.

Speaker C: As Alvaro Garcia puts it,

Speaker AA: Deepin is really good at uh, being able to crowdsource capex, being actually able for a project to leverage people's capital and people's effort to create a network without the central project having to invest all of that.

Speaker K: Right.

Speaker AA: It's also about being able to actually deploy and maintain a network that has locations that, that are, you know, in many different countries, very scattered and very distributed, uh, geographically around the earth. Because that's something that if you try to do from a company or a centralized perspective, it's very difficult, it's very expensive, you know, for example, for a company to build a network such as yield nets, where they have stations in almost every country on earth, they would have to set up uh, companies in every of those jurisdictions. They would have to hire people, pay leases. One of the key fundamentals is that uh, the more physically decentralized that a network needs to be, the more competitive advantage it has versus Web two players.

Speaker B: But why now? Well, we believe Deepin is on the cusp of finding strong market fit, driven by real world supply and demand dynamics that reinforce its flywheels. Until now, the demand side of Deepin has been subsidized by project treasuries and private capital, a strategy viable only until sustainable revenue streams emerge. But now we believe key factors are driving sustainable growth on the demand side, creating a sustainable flywheel effect that detaches Deepin from the broader crypto ecosystem.

Speaker C: As Alvaro Garcia puts it,

Speaker AA: for many Deepin projects, revenue has zero correlation with the crypto market. And that's unique as uh, Web3 category because if you think about it, DeFi NFTs, even exchanges, exchange tokens, layer ones, all of these, their fundamentals are proportional or let's say are highly correlated to the price of Bitcoin ultimately. So whenever we go into a bear market there's so much downward reflexivity or everything gets into a negative feedback loop. Now the thing about depin is that the fundamentals and the revenues from these projects have nothing to do with Web3 or the price of Bitcoin, which is very interesting because then what we can expect is that uh, if the fundamentals hold, at least there should be a bottom for the price of those tokens or the valuation of these protocols.

Speaker B: Historically, complexity has hindered Deepin adoption. But advancements in usability, regulatory clarity and AI for abstraction and coordination are reducing barriers to adoption. By abstracting the complexity, DBIM becomes a more attractive alternative to centralized solutions. And the second is pent up demand. Importantly, the growing need from AI for compute storage and connectivity, coupled with the ability to tap into idle resources through decentralized coordination is a game changer. With competitive pricing and decentralized supply that can absorb traditional bottlenecks, Deepin offers a compelling and scalable alternative. In increasingly strained infrastructure landscape. Deepin revolutionizes Blockchain's real world application by creating networks that deliver tangible services to everyday users. These networks leverage decentralized coordination to deploy infrastructure at a fraction of the cost and time required by traditional methods, unlocking scalable and efficient solutions for global infrastructure needs. This leads us to the post Web flywheel. The flywheel is a self reinforcing system where each component's growth amplifies the other, creating momentum and scalability over time. When resources like compute, storage, connectivity and latency are hosted in a decentralized way, they benefit from network effects where increased participation strengthens the network, translating into flywheels that drive adoption, efficiency and innovation. Compute, storage, connectivity and latency Deepin networks function as two or multi sided marketplaces where resources are offered in a decentralized way by a network of geographically distributed nodes operated by both providers and individuals. On the demand side, developers, businesses and users can tap into these resources for scalable, cost effective compute and storage solutions, leveraging the flexibility and redundancy of decentralized infrastructure. This transformation positions decentralized infrastructure as the backbone for the new Internet economy, providing the foundational resources necessary for AI agents, decentralized applications and other Post web technologies to thrive in an interconnected, highly automated and resilient ecosystem.

Speaker C: As Jasper Demeyer puts it, the Post

Speaker S: Web flywheel represents how, uh, decentralized and open source systems gain strengths through network effects, which is the underlying driver of all of this. So in these systems, the more participants and interactions there are, stronger the network becomes and it's this virtuous cycle that improves services, enhances efficiency and reduces risk by for example, uh, the reduced reliance on entities which can pose single points of failure. So by leveraging the principles of network effects in an open source decentralized system where a lot of services will be hosted by, um, third party nodes or like decentralized nodes, ultimately the flywheel is critically important.

Speaker B: As we conclude this chapter on the technology stack of the Post Web, it's clear that the past decade's innovations have laid the groundwork for a transformative new Internet architecture. The Post Web technology stack, spanning wallets, decentralized applications, DLT and personhood frameworks, is converging to create a cohesive ecosystem where users, agents and systems operate seamlessly. We hope by now it's clear that the Post Web represents not just an evolution, but a paradigm shift that will redefine how we interact with data systems and the economy itself. In the next episode, we'll examine the measurable impact of the Post Web on the digital and physical realms. We analyze how the technologies discussed will transform industries such as finance, healthcare media, consumer markets and more. Beyond technological shifts, we'll unpack the Post Web's new data economy and explore how agents orchestrate interactions between users, marketplaces and even among themselves in dynamic swarm like structures. We'll uncover how these advancements will reshape not just the Internet, but the very fabric of our economy and society. Post Web is no longer a distant vision. It is an imminent reality poised to reshape the Internet as we know it. Sam mhm.