Written by: a16z crypto
Compiled by: AididiaoJP, Foresight News
The economic model of the internet is changing. As the open internet increasingly shrinks into a prompt box, we can't help but wonder: Will AI bring about a more open internet, or a new maze of paywalls? Who will control it—large centralized companies or the vast user community?
This is precisely where cryptocurrency comes in. We have discussed the intersection with AI multiple times. In short, blockchain offers a new way to build internet services and networks—decentralized, credibly neutral, and user-owned. By reshaping the economic foundations of the current system, it can counterbalance the increasing centralization trend we see in the AI field, helping to build a more open and robust internet.
The idea that cryptocurrency can help build better AI systems, and vice versa, is not new, but it's often vaguely defined. Some combined areas—such as how to verify "human identity" in an era of low-cost AI proliferation—have already attracted developers and users. But other use cases seem years or even decades away from realization. Therefore, this article lists 11 use cases for the combination of AI and cryptocurrency, hoping to spark discussions on feasibility, unresolved challenges, and other issues. These ideas are all based on technologies being developed today, from processing massive micro-payments to ensuring humans own their relationships with future AI.
1. Let AI Remember You: Persistent Data and Interaction Context
Generative AI relies on data, but for many applications, context (i.e., the state and background information related to the interaction) is equally, if not more, critical.
Ideally, an AI system (whether an agent, a large language model interface, or another application) should remember your ongoing projects, your communication style, your preferred programming language, and many other details. But in reality, users often have to repeatedly rebuild this context across different sessions within the same application (like opening a new ChatGPT or Claude window) or even across different systems.
Currently, the context of a generative AI application is almost impossible to migrate to another application.
With blockchain, AI systems can transform key contextual elements into persistent digital assets. These assets can be loaded at the start of a session and seamlessly transferred between different AI platforms. Furthermore, given that interoperability and forward compatibility are fundamental properties of blockchain protocols, it might be the only solution to this problem and to establishing a commitment.
A natural application is AI-mediated games and media, where user preferences (from difficulty levels to key settings) can carry over across different games and environments. But the real value lies in knowledge-based applications, where AI needs to understand what the user knows and how they learn; and also in more specialized use cases, like programming. Of course, individual enterprises have already customized bots with global context for their businesses, but that context is usually not portable, not even between different AI systems within the same organization.
The closest current general solution is custom bots with fixed, persistent context. But portability of context between users within platforms has begun to appear in off-chain forms, such as the Poe platform allowing users to rent their custom bots to others.
Putting such activities on-chain would enable the AI systems we interact with to share a "context layer" that aggregates key information from all our digital activities. AI could instantly understand our preferences, thereby better optimizing the experience. Conversely, just like on-chain IP registration, enabling AI to reference persistent on-chain context also creates the possibility for innovative market interactions around prompts and information modules. For example, users could directly license or monetize their expertise while retaining data control. Of course, shared context will also enable many future applications we haven't yet imagined.
2. A Universal "Passport" for Agents
Identity is the canonical information that records "who or what," the "invisible plumbing" that underpins today's digital discovery, aggregation, and payment systems. Because platforms enclose these pipes within walls, the identity we experience is part of the finished product: Amazon assigns identifiers (ASIN or FNSKU) to products, lists them in one place, and helps users discover and pay. Facebook is similar: user identity is the foundation for its news feed and all in-app discovery features (including Facebook Marketplace, organic posts, and paid ads).
As AI agents develop, all this is about to change. The more companies use agents for customer service, logistics, payments, and other scenarios, the less their platforms resemble single-interface applications. Instead, agents will operate across multiple interfaces and platforms, accumulating deep contextual information and performing more tasks for users. But if an agent's identity is tied only to a single marketplace, it will be unusable in other important scenarios.
Therefore, agents need a single, portable "passport." Without it, there's no way to know how to pay the agent, verify its version, query its capabilities, understand who it works for, or track its reputation across different applications and platforms. An agent's identity needs to act as a wallet, an API registry, a changelog, and social proof, so that any interface (email, Slack, other agents) can identify and converse with it in the same way. Without "identity" as this shared foundational component, every integration requires rebuilding this plumbing from scratch, discovery becomes临时 and fragmented, and users lose context every time they switch channels or platforms.
We have an opportunity to redesign agent infrastructure from first principles. So, how do we build an identity layer that is richer than DNS records and credibly neutral? Agents should be able to accept payments, showcase capabilities, and exist in multiple ecosystems without fear of being locked into any specific platform, rather than rebuilding the "all-in-one" platform that bundles identity with discovery, aggregation, and payment. This is where the combination of cryptocurrency and AI is particularly useful, as blockchain networks provide permissionless composability, allowing developers to create more useful agents and better user experiences.
Overall, vertically integrated solutions like Facebook or Amazon currently have a better user experience—part of the complexity of building great products is ensuring the parts work together top-down. But this convenience comes at a high cost, especially as the software cost of building, marketing, monetizing, and distributing agents decreases and the coverage of agent applications expands. Matching the user experience of vertically integrated vendors requires effort, but a credibly neutral agent identity layer would allow entrepreneurs to own their "passport" and encourage more experimentation in distribution and design.
3. Forward-Compatible "Proof of Human"
As AI permeates various network interactions (including deepfakes and social media manipulation), it becomes increasingly difficult to determine whether the object of online interaction is a real person. This erosion of trust is not a future worry; it's already here: from comment armies on platform X to bots on dating apps, reality is blurring. In this environment, "proof of human" becomes critical infrastructure.
One way to prove human identity is through digital ID cards (including the centralized ones used by the TSA). Digital IDs contain all the information—usernames, PINs, passwords, third-party proofs (like nationality or credit status)—that can be used to verify identity. The value of decentralization here is obvious: when this data resides in centralized systems, the issuer can revoke access, charge fees, or facilitate surveillance at any time. Decentralization reverses this dynamic: the user, not the platform, controls their identity, making it more secure and censorship-resistant.
Unlike traditional identity systems, decentralized proof-of-human mechanisms (like Worldcoin's Proof of Human) allow users to control and manage their own identity and verify their human attributes in a privacy-preserving, credibly neutral manner. Just as a driver's license works anywhere, anytime it was issued, decentralized proof of human can also serve as a reusable base layer for any platform, including those not yet born. In other words, blockchain-based proof of human is forward-compatible because it provides:
- Portability: The protocol is a public standard that any platform can integrate. Decentralized proof of human can be managed by public infrastructure and controlled by the user. This makes it fully portable and compatible with any platform now or in the future.
- Permissionless Accessibility: Platforms can independently choose to recognize this proof of human ID without going through a "gatekeeper" API that might discriminate against different use cases.
The challenge in this area is adoption. Although we haven't seen many real application cases with scale effects yet, we expect that reaching a critical mass of users, a few early partners, and killer apps will accelerate its普及. Each application that adopts a specific digital identity standard makes that identity more valuable to users, thereby attracting more users to obtain that identity, which in turn makes the identity more attractive to more applications (as a means of verifying humans). Because on-chain IDs are designed to be interoperable, this network effect can grow rapidly.
We've already seen mainstream consumer applications and services in gaming, dating, and social media announce partnerships with World ID to help users confirm they are playing, chatting, and transacting with real people (and the specific real person they expect). This year has also seen new identity protocols, such as the Solana Attestation Service (SAS). Although SAS does not issue proof of human, it allows users to privately associate off-chain data (such as KYC checks required for compliance or investment eligibility certifications) with a Solana wallet to build a user's decentralized identity. All this suggests that the tipping point for decentralized proof of human may not be far away.
Proof of human is not just about banning bots; it's about drawing a clear line between AI agent and human networks. It allows users and applications to distinguish between human and machine interactions, opening up space to create better, safer, and more authentic digital experiences.
4. Decentralized Physical Infrastructure Networks for AI
AI is a digital service, but its development is increasingly constrained by physical infrastructure. Decentralized Physical Infrastructure Networks (DePIN) offer a new model for building and operating physical systems, helping to acquire the computing infrastructure that supports AI innovation, making it cheaper, more resilient, and more censorship-resistant.
How? The two biggest obstacles to AI development are energy and chip access. Decentralized energy can help provide more power, while developers are using DePIN to aggregate idle chips from sources like gaming PCs and data centers. These computers can together form a permissionless computing market, creating a level playing field for building new AI products.
Other use cases include distributed training and fine-tuning of large language models, and distributed networks for model inference. Decentralized training and inference could significantly reduce costs by leveraging otherwise idle computing resources. It can also provide censorship resistance, ensuring developers are not deplatformed by hyperscale cloud providers (large centralized cloud providers offering massively scalable computing infrastructure).
The concentration of AI models in the hands of a few companies is a persistent concern; decentralized networks can help create more cost-effective, censorship-resistant, and scalable AI.
5. Laying Tracks and Guardrails for AI-to-AI Interactions
As AI tools become better at solving complex tasks and executing multi-step interaction chains, AI will increasingly need to interact with other AIs without human intervention.
For example, an AI agent might need to request specific data related to computation, or recruit specialized AI agents for specific tasks (like assigning a stats bot to develop and run model simulations, or an image generation bot to participate in marketing material creation). AI agents will also create huge value in representing users to complete entire transaction flows or other activities, such as finding and booking flights based on preferences, or discovering and ordering new books from favorite genres.
Currently, there is no mature general market for inter-agent interactions. Most of these cross-platform queries happen only through explicit API connections or within closed ecosystems that support inter-agent calls.
More broadly, most AI agents today operate in isolated ecosystems, with relatively closed APIs and a general lack of architectural standardization. But blockchain technology can help establish open protocol standards, which is crucial for short-term adoption. In the long run, this also supports forward compatibility: as new types of AI agents evolve and are born, they are expected to connect to the same underlying network. Given their interoperable, open-source, decentralized, and usually easily upgradeable architecture, blockchains can more flexibly adapt to AI innovation.
As the market develops, several companies are already building blockchain "tracks" for inter-agent interactions. For example, Halliday recently launched its protocol, providing a standardized cross-chain architecture for AI workflows and interactions, with protocol-level protections to prevent AI from deviating from user intent. Meanwhile, companies like Catena, Skyfire, and Nevermind use blockchain to support direct payments between agents without human intervention. More such systems are under development, and Coinbase has even begun providing infrastructure support for such efforts.
6. Keeping AI/Vibe Applications in Sync
The recent generative AI revolution has made software development unprecedentedly easy. Coding speed has increased by orders of magnitude, and most importantly, it can be done in natural language, allowing even less experienced programmers to fork existing programs or create new ones from scratch.
But while AI-assisted coding brings new opportunities, it also introduces a lot of "entropy" (disorder) within and between programs. Vibe applications abstract away the complex network of underlying dependencies in software, but this can also make programs susceptible to functional and security flaws when inputs like source code repositories change. At the same time, when people use AI to create personalized applications and workflows, interfacing with others' systems becomes difficult. In fact, even two functionally identical vibe applications may have vastly different internal operations and output structures.
Historically, standardization efforts to ensure consistency and compatibility were first undertaken by file formats and operating systems, and more recently by shared software and API integrations. But in a world where software evolves, morphs, and branches in real-time, the standardization layer needs to be widely accessible and continuously upgradeable, while also maintaining user trust. Furthermore, AI alone cannot solve the problem of incentivizing people to build and maintain these links.
Blockchain provides an answer to both problems: a protocolized synchronization layer. It can be encapsulated into people's custom software and dynamically updated to ensure cross-platform compatibility during changes. In the past, a large enterprise might spend millions hiring "system integrators" like Deloitte to customize a Salesforce instance. Today, an engineer can create a custom interface to view sales information over a weekend. But as the number of custom software grows, developers will need help keeping these applications synchronized and running.
This is similar to the current development model of open-source software libraries, except it requires continuous updates rather than periodic releases, and needs an incentive layer. Both are easier to achieve with the help of cryptocurrency. Like other blockchain-based protocols, shared ownership of the synchronization layer incentivizes people to actively invest in improving it. Developers, users (or their AI agents), and other consumers can be rewarded for introducing, using, and developing new features and integrations.
In turn, shared ownership makes all users have a stake in the overall success or failure of the protocol, which effectively discourages bad behavior. Just as Microsoft is reluctant to break the .docx file standard because it harms users and the brand, co-owners of the synchronization layer are also reluctant to introduce clumsy or malicious code into the protocol.
Like all software standardization architectures we've seen, there is huge potential for network effects here. As the "Cambrian explosion" of AI-coded software continues, the network of heterogeneous, diverse systems that need to communicate with each other will expand dramatically. In short, "vibe coding" cannot stay in sync on "vibe" alone. Cryptocurrency is the answer.
7. Micropayments Supporting Revenue Sharing
AI agents and tools like ChatGPT, Claude, and Copilot promise new convenient ways to browse the digital world. But for better or worse, they are also shaking up the economic model of the open internet. We are already seeing the impact: for example, as students use AI tools more, traffic to educational platforms has significantly dropped; several US newspapers are suing OpenAI for copyright infringement. Without realigning incentives, we may face an increasingly closed internet: more paywalls, fewer content creators.
Of course, there are always policy solutions, but while legal processes advance, many technical solutions are emerging. Perhaps the most promising (and technically complex) solution is to build a revenue-sharing system into the web architecture. When AI-driven behavior facilitates a sale, the content source that informed that decision should receive a share. Affiliate marketing ecosystems already do similar attribution tracking and revenue sharing; a more complex version could automatically track and reward all contributors in the information chain. Blockchain clearly has a role to play in tracking the chain of provenance.
But such a system also requires new infrastructure with other characteristics: particularly micropayment systems capable of handling tiny transactions from multiple sources, attribution protocols that fairly assess different contributions, and governance models that ensure transparency and fairness. Many existing blockchain-based tools (like Rollups and L2s, AI-native financial institution Catena Labs, financial infrastructure protocol 0xSplits) show potential here, enabling near-zero-cost transactions and more granular payment splits.
Blockchain will enable complex agent payment systems through several mechanisms:
- Nanopayments can be distributed to multiple data providers: Through automated smart contracts, a single user interaction can trigger tiny payments flowing to all contributing sources.
- Smart contracts enable retroactive payments: After a transaction is completed, smart contracts can enforce retroactive payments, transparently and traceably compensating the information sources that contributed to the purchase decision.
- Enable complex programmable payment splits: Ensure revenue is distributed fairly through code-enforced rules rather than centralized decisions, establishing trustless financial relationships between autonomous agents.
As these emerging technologies mature, they can create a new economic model for media, capturing the complete value chain from creators to platforms to users.
8. Blockchain for Intellectual Property and Provenance Registration
Generative AI urgently needs efficient and programmable mechanisms to register and track intellectual property—both to confirm provenance and to pave the way for business models around IP access, sharing, and remixing. Existing IP frameworks, which rely on expensive intermediaries and ex-post enforcement, are no match for an era where AI consumes content instantly and generates new variations at the click of a button.
What we need are open, public registries that provide clear proof of ownership, which IP creators can interact with easily and efficiently, and which AI and other web applications can directly interface with. Blockchain is ideal because it enables IP registration without intermediaries, provides tamper-proof proof of provenance, and allows third-party applications to easily identify, license, and interact with that IP.
There is understandable skepticism about the overall idea that "technology can somehow protect IP," as the first two eras of the internet and the ongoing AI revolution have often been accompanied by a weakening of intellectual property protection. One problem is that many current IP-based business models focus on excluding derivative works rather than incentivizing and monetizing them. But programmable IP infrastructure not only allows creators, brands, and IP owners to assert their ownership in digital spaces, it also opens the door to business models around sharing IP in digital applications like generative AI.
We've already seen creators experiment with new models early in the NFT space, with companies using NFT assets on Ethereum to support network effects and value accumulation under CC0 brand building. More recently, we've seen infrastructure providers building protocols and even specialized blockchains (like Story Protocol) for standardized, composable IP registration and licensing. Some artists have begun using protocols like Alias, Neura, and Titles to license their styles and works for creative remixes. Meanwhile, Incention's Emergence series lets its fans co-create a sci-fi universe and characters, using a blockchain registry built on Story to record creative attribution.
9. Making Web Crawlers Compensate Content Creators
Today, the AI agents with the strongest product-market fit are not for coding or entertainment, but web crawlers, which autonomously browse the web, collect data, and decide which links to follow.
It is estimated that nearly half of internet traffic now originates from non-humans. Bots often ignore the norms of robots.txt (a file that should tell automated crawlers whether they are welcome, but has little actual authority) and use the scraped data to fortify the moats of some of the world's largest tech companies. Worse, websites ultimately foot the bill for these uninvited guests, providing bandwidth and computing resources to seemingly endless anonymous crawlers. In response, companies like Cloudflare and other content delivery networks offer blocking services, a patchwork solution that shouldn't need to exist.
We have pointed out that the original internet契约, the economic agreement between content creators and distribution platforms, is likely to collapse. Data is beginning to show this: over the past year, website owners have begun blocking AI crawlers en masse. In July 2024, only about 9% of the top 10,000 websites blocked AI crawlers; now that proportion is 37%. As more website operators upgrade their technology and user frustration continues, the proportion will only rise.
So, what if instead of paying CDNs to block all suspected bot traffic outright, we found a middle ground? AI bots could pay for data collection rights instead of freeloading on systems designed to guide human traffic. This is where blockchain comes in: in this scenario, each web crawler agent holds some cryptocurrency and negotiates on-chain with each website's "gatekeeper" agent or paywall protocol via something like x402.
Humans, meanwhile, could prove themselves on another channel via World ID and access content for free. This way, content creators and website owners could be compensated at the point of data collection for their contribution to large AI datasets, while humans could continue to enjoy an internet where information is meant to be free.
10. Privacy-Preserving, Tailored, and Non-Annoying Advertising
AI is already begun to influence how we shop online. But what if the ads we see every day were useful? There are many obvious reasons people dislike ads. Irrelevant ads are pure noise, and not all personalization is welcome. AI ads that are overly precise from vast consumer data can feel like an invasion of privacy. Other apps try to monetize content with unskippable ads.
Cryptocurrency can help solve some of these problems, offering an opportunity to重塑 the advertising model. Combined with blockchain, personalized AI agents can bridge the gap between "irrelevant ads" and "precision that feels intrusive," serving ads based on user-defined preferences. The key is they can do this without globally exposing user data, and can directly compensate users for sharing data or interacting with ads.
This requires some technical requirements:
- Low-fee digital payments: To compensate users for ad interactions (views, clicks, conversions), companies need to make small, high-frequency payments. To operate at scale, we need fast, high-throughput systems with very low fees.
- Privacy-preserving data verification: AI agents need to be able to prove that a consumer meets certain demographic attributes. Zero-knowledge proofs can verify such attributes while preserving privacy.
- Incentive mechanisms: If the internet adopts a micropayment-based monetization model (e.g., less than $0.05 per interaction), users will have the choice to watch ads in exchange for small payments, thus transforming the current model from "extractive" to "participatory."
People have been trying to make online ads relevant for decades, but rethinking advertising through the lens of cryptocurrency and AI could finally make ads more useful. Make them tailored but not annoying, and in a way that benefits all parties: for builders and advertisers, it unlocks new incentive structures that are more sustainable and consistent; for users, it provides more ways to discover and browse the digital world.
All this will make the advertising space more valuable, not less. It might also replace today's entrenched, extractive ad economy with a more humane system: one that treats users as participants, not products.
11. AI Companions Owned and Controlled by Humans
Many people spend more time on their devices than in face-to-face interactions, and increasingly that time is spent interacting with AI models and AI-curated content. These models already provide a form of companionship, whether for entertainment, providing information, catering to niche interests, or educating children. It's not hard to imagine that in the near future, AI-based educational, medical, legal advisory, and friendship companions will become popular modes of human interaction.
Future AI companions will have infinite patience and be tailored to specific individuals and their use cases. They will not just be assistants or robotic servants; they may become cherished relationships. Therefore, who owns and controls these relationships—the user or companies and other intermediaries—becomes equally important. If you've been concerned about content control and censorship on social media over the past decade, this issue will become exponentially more complex and personal in the future.
Arguments have already been made that censorship-resistant hosting platforms like blockchain are the clearest path to user-controlled, uncensorable AI. Admittedly, individuals can run on-device models and buy their own GPUs, but most can't afford it or don't know how.
Although widespread adoption of AI companions is still some time away, related technology is advancing rapidly: text-based, seemingly human companions are already quite good; visual avatars have improved significantly; blockchain performance continues to improve. To ensure uncensorable companions are easy to use, we need better user experiences from cryptocurrency applications. Thankfully, wallets like Phantom have made interacting with blockchains much simpler, and embedded wallets, passkeys, and account abstraction allow users to hold self-custody wallets without managing complex seed phrases. Technologies like high-throughput trustless computers utilizing optimistic and zero-knowledge coprocessors will also make it possible to form meaningful and lasting relationships with digital companions.
In the near future, the focus of discussion will shift from "when will we see lifelike digital companions and avatars" to "who and what will be able to control them."
