Author: a16z
Compiled by: Felix, PANews
The economics of the internet are changing. As the open web degenerates into a search box, it’s worth asking: Will AI lead to an open internet, or a new maze of paywalls? And who will control it, large centralized companies or a broad user base?
This is where cryptocurrencies come into play. The intersection of AI and cryptocurrencies has been explored many times before; in short, blockchains are a new way to build internet services and networks that are decentralized, trusted, neutral, and owned by their users. They balance many of the centralized forces we’ve seen in AI systems by renegotiating the economic mechanisms that underpin today’s systems, helping to enable a more open, powerful internet.
The idea that cryptocurrencies can help build better AI systems, and vice versa, is not new, but the definitions are often unclear. Some intersections—like verifying “proof of human identity” amid the proliferation of low-cost AI systems—are already attracting the attention of builders and users. But other use cases seem years or even decades away. So this post shares 11 use cases at the intersection of cryptocurrencies and AI to start a discussion about future possibilities, challenges, and more. These use cases are based on technology being built today, from processing massive amounts of small payments to ensuring that humans are in control of their relationship with future AI.
1. Persistent data and context in AI interactions
Generative AI relies on data, but for many applications, context (state and background information related to the interaction) is equally important or even more important.
Ideally, an AI system, whether it’s an agent, LLM interface, or other application, would remember details like the type of project you’re working on, your communication style, and your preferred programming language. But in practice, users often need to re-establish this context across different interactions within a single application (such as every time they launch a new ChatGPT or Claude command-line interface), not to mention when switching between different systems.
Currently, the context in one generative AI application is almost never, or even impossible, to transfer to other applications.
With blockchain technology, AI systems can keep critical contextual information in the form of persistent digital assets that can be loaded at the start of a session and seamlessly transferred between different AI platforms. In addition, given the characteristics of blockchain protocols, blockchain may be the only solution to this problem that is both forward-compatible and ensures interoperability.
A natural application for this technology is AI-assisted gaming and media, where user preferences (from difficulty level to button settings) can be consistent across different games and environments. But its real value lies in knowledge applications. In these applications, AI needs to understand what users already know and how they learn; there are also more specialized application scenarios, such as programming. Of course, some companies have developed personalized bots and operate based on the global context of a specific business - but in this case, the context is usually not portable.
The closest general solution we’ve seen so far is custom bots with fixed, persistent context. However, portability of context between users within a platform is starting to emerge off-chain; for example, through Poe, users can rent out their custom bots to others.
Bringing this type of activity on-chain enables AI systems to share a contextual layer that is comprised of the key elements of all digital activity. They can instantly learn preferences and better tailor and optimize experiences. In turn, like an on-chain intellectual property registry, enabling AI to reference persistent on-chain context creates the possibility for new market interactions around tips and information modules — for example, users can directly license or monetize their expertise while retaining control of their data. Shared context will make many things possible that have not yet been conceived.
2. General identity of the agent
Identity is the authoritative record of who something or someone is, and it underpins today’s digital discovery, aggregation, and payment systems. Because platforms hide this underpinning behind the scenes, the identity you experience is only part of the finished product: Amazon assigns an identifier (ASIN or FNSKU) to a product, places it in a central location, and helps users discover and pay for it. The same is true for Facebook: a user’s identity is the foundation of their News Feed and discovery throughout the app, including Facebook Marketplace listings, organic posts, and paid ads.
All that is about to change as AI agent technology advances. As more companies use agents to handle customer service, logistics, payments, and other use cases, their platforms will no longer look like single-interface applications. Instead, they will exist across multiple interfaces and platforms, accumulating rich context and performing more tasks for users. But tying an agent’s identity to just one market makes it unusable in other important places, like email conversations, Slack channels, and other products.
That’s why proxies need a single, portable “passport.” Without a passport, there’s no way to understand how to pay a proxy, verify its version, query its capabilities, understand who the proxy represents, or track its reputation across apps and platforms. The proxy’s identity needs to act as a wallet, an API registry, a change log, and social proof — so that any interface (email, Slack, or another proxy) can resolve and communicate with it in the same way. Without a shared “identity” primitive, every integration needs to be rebuilt from scratch, discovery mechanisms remain ad hoc, and users lose contextual information every time they switch channels or platforms.
Designing proxy infrastructure from scratch is possible. So how do you build a trusted neutral identity layer that’s richer than DNS records? Rather than redesigning a monolithic platform that combines identity with discovery, aggregation, and payments, it’s possible to enable proxies to accept payments, list functionality, and exist in multiple ecosystems without worrying about being locked into a specific platform. This is where crypto meets AI, as blockchain networks offer permissionless composability that enables builders to create more useful proxies and better user experiences.
In general, vertically integrated solutions like Facebook or Amazon currently have a better user experience - part of the inherent complexity of building great products is making sure the pieces fit together properly from top to bottom. But this convenience comes at a high price, especially as the cost of building software for aggregation, marketing, monetization, and distribution agents decreases, and the applicability of agent applications continues to expand. There is still work to be done to match the user experience of vertically integrated providers, but a trusted, neutral agent identity layer would enable entrepreneurs to own their own passports - and encourage innovation in distribution and design.
3. Forward-compatible identity proof
As AI becomes more prevalent, it’s becoming increasingly difficult to determine if the person you’re communicating with online is actually a real person. This erosion of trust isn’t a future worry; it’s already here. From the army of reviewers on Platform X to bots on dating apps, reality is starting to blur. In this environment, proof of identity becomes critical infrastructure.
One way to prove you’re human is through a digital identity. A digital identity encompasses everything a person can use to verify their identity — usernames, PINs, passwords, third-party proofs (such as citizenship or creditworthiness), and other credentials. The value of decentralization is clear here: when this data lives in a centralized system, the issuer can revoke access, charge fees, or facilitate surveillance. Decentralization flips this dynamic: users take control of their own identity, making it more secure and less susceptible to censorship.
Unlike traditional identity systems, decentralized identity proof mechanisms (such as World ID's "Proof of Humanity") allow users to control and maintain their own identity and verify their human identity in a privacy-preserving and trustworthy neutral way. And just like a driver's license can be used everywhere regardless of when and where it was issued, decentralized identity proof can also serve as a reusable underlying foundation for any platform (including those that have not yet appeared). In other words, blockchain-based identity proof is forward-compatible because it provides:
Portability: The protocol is an open standard that any platform can integrate. Decentralized PoPs are managed through a public infrastructure and are fully controlled by users. This makes it fully portable and any platform now or in the future will be compatible with it.
Permissionless accessibility: Platforms can choose to identify PoP IDs autonomously without going through a gateway API that may discriminate between different use cases.
The challenge in this space is adoption: while we haven’t seen real use cases for proving identity at scale yet, we expect adoption to accelerate as the number of users reaches a certain scale, early partners emerge, and killer apps are launched. Each application that leverages a particular digital identity standard makes that identity type more valuable to users; this drives more users to acquire that identity; which in turn makes that identity more attractive to applications as a way to authenticate identity. (And because on-chain identity is designed to be interoperable, these network effects can scale rapidly)
We’ve already seen mainstream consumer apps and services in the gaming, dating, and social media sectors announce partnerships with World ID to help people confirm they’re interacting with real people — indeed, the people they expect. This year has also seen the emergence of new identity protocols, including the Solana Authentication Service (SAS). Although not an issuer of identity, SAS allows users to privately associate off-chain data (such as compliance KYC checks or investment certification status) with a Solana wallet to build a user’s decentralized identity. All of this suggests that a tipping point for decentralized identity may not be far away.
Proof of identity isn’t just about banning bots, it’s about creating a clear boundary between AI agents and human networks. It enables users and applications to differentiate between human and machine interactions, creating space for better, safer, and more authentic digital experiences.
4. DePIN with AI
AI may be a digital service, but its development is increasingly constrained by physical infrastructure. Decentralized Physical Infrastructure Networks (DePINs)—a new model for building and operating real-world systems—can help make the computing infrastructure that AI innovation relies on more accessible, cheaper, more resilient, and harder to censor.
How is it done? The two biggest barriers to AI development have always been energy and chip access. Decentralized energy helps provide more power, but developers are also using DePIN to pool idle chips from gaming PCs, data centers, and other sources. These computers can be pooled to form a permissionless computing market, creating a level playing field for developing new AI products.
Other use cases include distributed training and fine-tuning of LLMs, and distributed networks for model inference. Decentralized training and inference can significantly reduce costs because they utilize otherwise idle computing resources. In addition, they can provide censorship resistance, ensuring that developers will not be banned by hyperscale cloud service providers.
The concentration of control over AI models by a few companies has long been a concern; decentralized networks can help create AI that is more cost-effective, censorship-resistant, and scalable.
5. Infrastructure and safeguards for interactions between AI agents, end service providers, and users
As AI tools become more capable of solving complex tasks and executing multi-layered chains of interactions, AI will increasingly need to interact with other AI without the intervention of human controllers.
For example, an AI agent might need to request specific data related to a computation, or recruit specialized AI agents to perform a specific task. AI agents will also create tremendous value by completing entire transaction flows or any other activities on behalf of users—like finding and booking a flight based on someone’s preferences, or discovering and ordering a new book in their favorite genre.
There is currently no mature, general agent-to-agent market - this type of cross-query is mostly only possible through API connections, or within an AI agent ecosystem that maintains inter-agent calls as an internal function.
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 standards for protocols, which is critical for adoption in the short term. In the long term, this also helps with forward compatibility: as new types of AI agents continue to evolve and emerge, they can expect to be able to access the same underlying network. Given blockchains’ interoperability, open source, decentralized, and generally more easily upgradeable architectures, they can more easily adapt to a variety of new AI innovations.
As the market develops, many companies are already building blockchain infrastructure for agent-to-agent interactions: Halliday, for example, recently launched its protocol to provide a standardized cross-chain architecture for AI workflows and interactions. Meanwhile, Catena, Skyfire, and Nevermind are using blockchain to support payments between AI agents without human intervention. Many more such systems are in development, and Coinbase has even begun providing infrastructure support for these efforts.
6. Keep AI/ambience encoding apps in sync
The recent revolution in generative AI has made it easier than ever to build software. Coding is orders of magnitude faster and can be done in natural language, so even inexperienced programmers can fork existing programs and build new ones from scratch.
But while AI-assisted programming opens up these novel opportunities, it also introduces a great deal of uncertainty within and between programs. “Ambient programming” abstracts away the complex dependencies behind software—but this can also make programs susceptible to functional and security flaws when source code repositories and other inputs change. At the same time, as people leverage AI to create personalized applications and workflows, their interactions with other systems can become more difficult. In fact, even two ambient-programmed programs that perform the same task can have very different operations and output structures.
Historically, standardization efforts to ensure consistency and compatibility were initially accomplished by file formats and operating systems, and more recently by shared software and API integrations. But in a world where software evolves, changes, and branches in real time, the standardization layer needs to be widely accessible and continually upgradeable—all while maintaining user trust. Furthermore, AI alone cannot solve the problem of incentivizing people to make and maintain these connections.
Blockchain solves both problems at once: protocolized synchronization layers that are wrapped into people’s custom software builds and updated dynamically. Today, an engineer can create a custom interface to view sales information in a weekend, but as the amount of custom software grows, developers will need help keeping these applications in sync and functioning properly.
This is similar to how open source software libraries are developed today, but with continuous updates rather than periodic releases, and with incentives. Both of these are much easier to achieve with the help of cryptography. Just like other blockchain-based protocols, shared ownership of the sync layer incentivizes people to actively invest in improving it. Developers, users (and/or their AI agents), and other consumers can be rewarded for introducing, using, and improving new features and integrations.
In turn, shared ownership gives all users a stake in the overall success of the protocol, which can act as a buffer against bad behavior. Just as Microsoft was reluctant to break the .docx file standard because of the ripple effects it would have on its users and brand, the co-owners of the sync layer are reluctant to introduce clumsy or malicious code into the protocol.
There is huge potential for network effects here. As the Cambrian explosion of AI coding software continues, the network of heterogeneous, diverse systems that need to communicate with each other will expand dramatically. In short: Ambient coding requires more than just atmosphere to stay in sync. Cryptography is the answer.
7. Support micropayments for revenue sharing
AI agents and tools like ChatGPT, Claude, and Copilot promise to give people a convenient new way to explore the digital world. But for better or worse, they’re also shaking the economic foundations of the open internet. We’re already seeing this happening — for example, education platforms have seen a significant drop in traffic as students increasingly use AI tools, and several US newspapers are suing OpenAI for copyright infringement. Without a realignment of incentives, we may see an increasingly closed internet with more paywalls and fewer content creators.
Of course, there are always policy solutions, but in the process of implementation, some technical solutions have also emerged. Perhaps the most promising (and technically complex) solution is to embed revenue sharing mechanisms into the network structure. When AI-driven behavior leads to a transaction, the source of the content that provided information for that decision should receive a portion of the revenue. The affiliate marketing ecosystem already does this kind of attribution tracking and revenue sharing; a more complex version can automatically track and reward all contributors in the chain of information. Blockchain can obviously play a role in tracking this chain of provenance.
But such a system will also require new infrastructure with other capabilities—particularly micropayment systems that can handle tiny transactions from many sources, vesting protocols that fairly value different types of contributions, and governance models that ensure transparency and fairness. Many existing blockchain-based tools (such as rollups and L2 solutions, AI-native financial institution Catena Labs, and financial infrastructure protocol 0xSplits) show potential here, enabling near-zero-cost transactions and more granular payment allocations.
Blockchain will enable complex proxy payment systems through a variety of mechanisms:
Nanopayments can be distributed across multiple data providers, and through automated smart contracts, a single user interaction can trigger micropayments to all contributing sources.
Smart contracts enable retroactive payments, compensating the sources of information that contributed to the purchasing decision in a fully transparent and traceable manner after the transaction is completed.
In addition, blockchain enables complex and programmable payment distribution, ensuring fair distribution of income through code-enforced rules rather than centralized decision-making, thereby establishing trustless financial relationships between autonomous agents.
As these emerging technologies mature, they can create a new economic model for media that captures the entire value creation chain from creators to platforms to users.
8. Blockchain as a registry of intellectual property and provenance
Generative AI urgently needs efficient and programmable mechanisms to register and track IP—both for the purpose of determining provenance and to support business models around access, sharing, and re-creation of IP. Existing IP frameworks—which rely on costly intermediaries and ex post enforcement—are ill-equipped to cope with an era where AI consumes content instantly and can generate new variations with a single click.
As these emerging technologies mature, they can create a new economic model for media that captures the entire value creation chain from creators to platforms to users.
What is needed is an open public registry that provides clear proof of ownership, that IP creators can interact with easily and efficiently, and that AI and other web applications can interface directly with. Blockchains are well suited for this purpose because they make it possible to register IP without relying on middlemen and provide immutable proof of origin; they also make it easy for third-party applications to identify, authorize, and interact with that IP.
In the early days of the NFT space, creators began experimenting with new models, with some companies leveraging NFT assets on Ethereum to support network effects and value accumulation, and adopting CC0 for brand building. Recently, infrastructure providers are building protocols and even specialized blockchains (such as Story Protocol) to enable standardized and composable intellectual property registration and licensing. Some artists have begun using these tools to license their styles and works through protocols such as Alias, Neura, and Titles. Meanwhile, Incention's Emergence series allows its fan base to participate in the co-creation of a sci-fi universe and its characters, and has established a blockchain registry on Story to record who created what.
9. Web scrapers that help content creators get paid
Today, the AI agents with the best product-market fit aren’t for programming or entertainment, but rather web crawlers — agents that autonomously browse the web, collect data, and decide which links to follow. These crawlers help compensate content creators.
It’s estimated that about half of all internet traffic now comes from non-human sources. Bots often ignore the rules of robots.txt files (which are supposed to tell automated web crawlers whether they’re welcome to visit a site, but actually have little authority), and use the data they extract to bolster the defenses of some of the world’s largest tech companies. Worse, sites end up paying for these unwanted guests, providing bandwidth and CPU resources to an endless stream of anonymous crawlers. To that end, companies like Cloudflare and other CDNs (content delivery networks) offer blocking services. It’s a patchwork service that shouldn’t exist.
I’ve pointed out before that the original contract of the internet — the economic contract between content creators and distribution platforms — is likely to break down. So what if, instead of paying a CDN to outright block any website that looks like a robot, we find a compromise? Instead of getting it “for free” from a system designed to drive traffic to a website, AI bots could pay for the right to collect data. This is where blockchain comes in: in this scenario, each web crawler agent holds some cryptocurrency and negotiates on-chain with each website’s “bouncer” agent or paywall agreement via x402. (The challenge, of course, is that the robots.txt system, also known as the Robots Exclusion Standard, has been ingrained in the way internet companies do business since the 1990s. Solving this would require large-scale group coordination, or the involvement of a CDN like Cloudflare).
But there is another way to get free content by proving your identity with a “World ID” (see above). This way, content creators and website owners can be compensated when they collect large AI datasets, and humans can continue to enjoy an internet where information craves freedom.
10. Personalized advertising that protects privacy
AI is already starting to impact the way we shop online, but what if the ads we see every day were… helpful? It’s easy to see why people don’t like ads. Unrealistic ads are pure noise. At the same time, not all personalization is created equal. Overly accurate AI ads can feel like an invasion of privacy. Other apps try to monetize by placing content behind unskippable ads (like streaming content services or game levels).
Cryptocurrencies can help solve some of these problems, providing an opportunity to reimagine how advertising works. Combined with blockchain, personalized AI agents can bridge the gap between irrelevant and disturbing ads, serving ads based on user-defined preferences. But more importantly, they can do this without exposing user data globally, while directly compensating users for sharing their data or interacting with ads.
Some of the technical requirements here include:
Low-fee digital payments: To compensate users for ad interactions (views, clicks, conversions), companies need to send small, frequent payments. To achieve scale, a fast, high-throughput, low-fee system is needed.
Privacy-preserving data verification: AI agents need to be able to prove that consumers meet certain demographic characteristics. Zero-knowledge proofs can verify demographic characteristics while preserving privacy.
Incentive Model: If the internet adopted a micropayment-based monetization approach (e.g., less than $0.05 per interaction as described above), users would be able to opt-in to ads in exchange for micropayments, thus transforming the current model from an extraction model to a participation model.
Rethinking advertising through the lens of crypto and AI can ultimately make advertising more useful. Ads that are tailored, but not creepy, and in a way that benefits everyone: For content creators and advertisers, it opens up new incentives that are more sustainable and better aligned. For users, it provides more ways to explore and interact with the digital world.
All of this will make advertising more valuable. It could also upend today’s entrenched, extractive ad economy and replace it with a more human-centric system: one that treats users as participants rather than products.
11. AI companions, owned and controlled by humans
Many people spend more time on their devices than in face-to-face interactions, especially with AI models and AI-curated content. All of these modes already provide a form of companionship, whether it is entertainment, information, indulging in a niche interest, or educating children. It is not difficult to imagine that in the near future, AI-based education, medical care, legal advice, and friendship companions will become a popular mode of human interaction.
The AI companions of the future will be infinitely patient and tailored to specific individuals and use cases. Beyond acting as helpers or robot servants, they can become highly valuable relationships. As a result, the question of who will have ownership and control over these relationships (whether it’s the user, a company, or another intermediary) becomes equally important. If you’ve already worried about the management and censorship of social media over the past decade, this question will become more complex and more personal in the future.
Censorship-resistant, hosted platforms like blockchains offer the most compelling path to censorship-resistant, user-controlled AI. While it’s true that individuals can run models on-device and buy their own GPUs, most people either can’t afford them or simply don’t know how to do so.
While we’re still a long way from widely available AI companions, all of these technologies are improving rapidly: Human-looking text-based companions are already excellent. Visual imagery has also improved significantly. Blockchain performance is also getting better. To ensure that censorship-resistant companions are easy to use, better user experiences need to be provided for crypto-powered applications. Thankfully, wallets like Phantom make interacting with the blockchain much simpler, while embedded wallet, key, and account abstractions enable users to hold self-hosted wallets without the complexity of storing seed phrases themselves. Technologies such as high-throughput trustless computers using OP coprocessors and zero-knowledge proof coprocessors will also make it possible to build meaningful and lasting relationships with digital companions.
In the near future, the conversation will shift from when we’ll see lifelike digital companions and avatars to who and what will be able to control them.
Related reading: AI×Crypto intersection: in-depth analysis of five major AI Layer1 projects
