a16z: 11 Use Cases for the Convergence of Crypto and AI
TechFlow Selected TechFlow Selected
a16z: 11 Use Cases for the Convergence of Crypto and AI
When AI redefines digital interaction and Crypto reshapes value distribution, their convergence gives rise to 11 technical integration scenarios—from data persistence management to decentralized identity verification—exploring the technological frameworks and ecosystem possibilities of the open web.
Navigating Web3 tides with focused insightsAuthors: Scott Duke Kominers, Sam Broner, Jay Drain, Guy Wuollet, Elizabeth Harkavy, Carra Wu, Matt Gleason
Translation: Saoirse, Foresight News
The economic logic of the internet has quietly shifted. As open networks gradually contract into a single "prompt bar," we must ask: Will artificial intelligence lead us toward an open internet, or trap us in a maze of new paywalls? Who will hold control—the large centralized corporations, or the broader user base?
This is where crypto comes in. We've discussed the intersection of AI and crypto many times before. In short, blockchain represents a new paradigm for restructuring internet services—one that enables decentralized, credibly neutral, and user-owned networks. By redefining the economic rules underpinning existing systems, blockchain offers a viable path to counterbalance centralization in AI, paving the way for a more open and resilient internet ecosystem.
The idea of mutual empowerment between crypto and AI systems is not new, but how they integrate remains poorly defined. Some crossover areas—like verifying "human identity" amid an explosion of low-cost AI tools—have captured developer and user attention, while other applications may take years or even decades to materialize. This article outlines 11 cross-application scenarios at the intersection of AI and crypto to advance the discussion: exploring potential synergies, identifying challenges, and envisioning future innovations. These scenarios are grounded in current technological capabilities, spanning fields from micro-payment processing to ensuring human primacy in future AI interactions.
1. Persistent Data and Context in AI Interactions
Author: Scott Duke Kominers
Generative AI thrives on data, but in many use cases, context—state and background information related to interactions—is equally, if not more, critical.
Ideally, AI systems such as agents, large language model interfaces, or other applications would remember details like a user’s work type, communication style, preferred programming languages, and more. In reality, users often have to re-establish this context across different sessions within the same app (e.g., starting a new ChatGPT or Claude session), let alone when switching between systems. Currently, context from one generative AI application rarely transfers to another.
Blockchain technology can transform key contextual elements into persistent digital assets, enabling them to be loaded at the start of a session and seamlessly transferred across different AI platforms. Moreover, due to its inherent properties, blockchain may be the only solution capable of simultaneously achieving forward compatibility and interoperability.
This use case is especially valuable in AI-mediated gaming and media—user preferences (from difficulty settings to key bindings) could remain consistent across games and environments. But the real value lies in knowledge-intensive applications (where AI needs to understand a user's knowledge base and learning patterns) and specialized AI tools (such as coding assistants). While some companies have developed custom bots for specific business contexts, these typically lack cross-system portability—even within different AI tools inside the same organization.
Institutions are only beginning to recognize this issue. Current general-purpose solutions involve custom bots with fixed contexts. However, off-chain context migration between users is already emerging—for example, on Poe, users can rent out their customized bots to others.
Bringing such scenarios on-chain would allow our interacting AI systems to share a unified context layer containing key elements of all digital activity. They would instantly understand user preferences and optimize experiences accordingly. Similarly, on-chain intellectual property registries could enable AI to reference persistent, on-chain context, opening new market-driven interactions around prompts and information modules—users could directly license or monetize their expertise while retaining data control. Shared context will also unlock possibilities we haven’t yet imagined.
2. Universal Identity System for Agents
Author: Sam Broner
Identity—the authoritative record of “what something is”—is the foundational architecture behind today’s digital discovery, aggregation, and payment systems. Because platforms keep this architecture locked within walled gardens, identity has become just another product feature: Amazon assigns unique identifiers (ASINs or FNSKUs) to products, aggregates listings, and facilitates discovery and payments; Facebook treats user identity as the core of its feed and in-app discovery engine, encompassing product listings, native posts, and paid ads.
With the evolution of AI agents, this status quo is about to change. As more businesses adopt agents—for customer service, logistics, payments, etc.—these agents won't be confined to a single interface. Instead, they’ll operate across multiple ecosystems, accumulate deep context, and perform increasingly complex tasks. But if agent identity is tied to just one marketplace, it becomes unusable in other critical contexts like email threads, Slack channels, or other products.
Hence, agents need a single, portable “digital passport.” Without it, there’s no reliable way to pay agents, verify their versions, query their capabilities, identify who they serve, or track their reputation across apps and platforms. An agent’s identity must function as a wallet, API registry, changelog, and social proof—all in one—to ensure any interface (email, Slack, or another agent) can parse and interact with it consistently. Without shared identity information, every system integration requires rebuilding foundational infrastructure from scratch. Discovery mechanisms remain ad hoc, and users lose context whenever they switch channels or platforms.
We now have the chance to design agent infrastructure from first principles. How do we build a credible, neutral identity layer superior to DNS records? Agents should avoid repeating the monolithic platform mistake of bundling identity with discovery, aggregation, and payments. Instead, they should be able to receive payments, showcase functionality, and operate across multiple ecosystems without fear of lock-in. This is precisely where the convergence of crypto and AI adds value—blockchain networks offer permissionless composability, empowering developers to build more useful agents and better user experiences.
Currently, vertically integrated solutions (like Facebook or Amazon) deliver superior UX. One inherent challenge in building great products is ensuring seamless top-down coordination. But this convenience comes at a high cost—especially as the cost of building agent aggregation, marketing, monetization, and distribution software declines, and agent use expands. While matching the UX quality of vertical platforms remains difficult, creating a credible, neutral identity layer for agents will empower entrepreneurs to own their “digital passports” and explore innovative distribution and design strategies.
3. Forward-Compatible Proof-of-Human Mechanisms
Authors: Jay Drain Jr., Scott Duke Kominers
As AI permeates online interactions—including deepfakes and social media manipulation—it’s becoming harder to determine whether we’re communicating with real humans. This breakdown in trust is already underway—from bot armies on X (formerly Twitter) to fake profiles on dating apps. The boundary between virtual and real is blurring. In this environment, proof-of-human identity becomes core digital infrastructure.
One way to prove “I am human” is through digital IDs—including centralized ones like those used by the TSA. Digital ID encompasses usernames, PINs, passwords, third-party attestations (like citizenship or credit scores)—anything used to authenticate identity. Decentralization brings clear advantages: when data lives in centralized systems, issuers can revoke access, charge fees, or surveil users. Decentralized models reverse this power dynamic—users, not platform operators, control their identities, making them more secure and censorship-resistant.
Unlike traditional identity systems, decentralized proof-of-human mechanisms (like World’s Proof of Humanity) let users manage their own identity data and verify humanity in privacy-preserving, credibly neutral ways. Just as a driver’s license works anywhere regardless of where it was issued, decentralized proof-of-human can serve as a universal protocol across platforms—even those not yet invented. In other words, blockchain-based proof-of-human offers forward compatibility thanks to two key benefits:
-
Portability: These protocols are public standards. Any platform can integrate them. Decentralized proof-of-human can be managed via public infrastructure, fully controlled by users, and universally compatible—now and in the future.
-
Permissionless Access: Platforms can choose to recognize a proof-of-human ID without relying on APIs that might discriminate across use cases.
The challenge lies in adoption: despite growing interest, no large-scale proof-of-human use cases exist yet. However, we expect adoption to accelerate once user thresholds are reached, early partnerships form, and killer apps emerge. Each new app adopting a specific digital ID standard increases that ID’s value to users, encouraging more people to obtain one—a positive feedback loop. And because on-chain IDs are designed for interoperability, network effects can compound rapidly.
We’re already seeing mainstream consumer apps in gaming, dating, and social media announce partnerships with World ID to help users confirm they’re interacting with real people, not bots. New identity protocols like Solana Attestation Service (SAS) have also emerged—though SAS isn’t a proof-of-human issuer, it allows users to privately link off-chain data (like KYC compliance or investment qualifications) to Solana wallets, helping build decentralized identity. All signs suggest the tipping point for decentralized proof-of-human may be near.
Proof-of-human isn’t just about blocking bots—it’s about drawing a clear line between human networks and AI agents. It empowers users and apps to distinguish between human and machine interactions, enabling higher-quality, safer, and more authentic digital experiences.
4. Decentralized Infrastructure (DePIN) for AI
Author: Guy Wuollet
AI is a digital service, but its growth is increasingly constrained by physical infrastructure. Decentralized Physical Infrastructure Networks (DePIN) offer a new model for building and operating real-world systems, democratizing the computational infrastructure behind AI innovation to make it more affordable, resilient, and censorship-resistant.
How can this be achieved? Two major bottlenecks in AI development are computing supply and chip access. Decentralized compute networks can increase available processing power, with developers leveraging DePIN to aggregate idle chips from sources like gaming PCs and data centers. These devices can form permissionless compute markets, leveling the playing field for building new AI products.
Other applications include distributed training and fine-tuning of large language models, and decentralized inference networks. Distributed training and inference—by utilizing otherwise idle compute resources—can drastically reduce costs while offering censorship resistance, ensuring developers aren’t shut down by hyperscale cloud providers (like dominant centralized cloud giants).
A long-standing issue has been the concentration of AI models among a few firms. Decentralized networks help build a more economical, censorship-resistant, and scalable AI ecosystem.
5. Infrastructure and Rule Frameworks for Interaction Between AI Agents, End-Service Providers, and Users
Author: Scott Duke Kominers
As AI tools grow more capable at solving complex tasks and executing multi-step interaction chains, AI systems will increasingly need to interact with each other autonomously—without human intervention.
For instance, an AI agent might request specific data related to a computational task, or recruit specialized agents to complete subtasks—such as assigning a statistical bot to develop and run simulation models, or calling on an image-generation bot when creating marketing materials. Agents can also create significant value by handling end-to-end transactions or activities on behalf of users—like finding and booking flights based on personal preferences, or discovering and ordering new books of a favorite genre.
Today, there is no mature, general-purpose market for agent-to-agent interaction. Most cross-system queries rely on explicit APIs or occur only within AI ecosystems that internally support agent calling.
Most AI agents currently operate in siloed environments, with closed, non-standardized APIs. Blockchain technology, however, can help establish open standards—critical for short-term deployment and long-term forward compatibility. As new AI agents evolve and emerge, they can plug into the same underlying network. Blockchain’s interoperable, open-source, decentralized, and easily upgradable architecture makes it highly adaptable to the fast-moving pace of AI innovation.
As the market develops, several companies are already building blockchain infrastructure for agent interaction: Halliday recently launched a protocol providing standardized cross-chain architecture for AI workflows and interactions, including safeguards at the protocol level to ensure AI actions stay within user intent. Catena, Skyfire, and Nevermind use blockchain to enable automatic payments between AI agents without human involvement. More such systems are in development, and Coinbase has begun supporting these efforts with infrastructure.
6. Ensuring Synchronization of AI / Vibe-Coded Applications
Authors: Sam Broner, Scott Duke Kominers
Generative AI has revolutionized software development, boosting coding speed by orders of magnitude—and most importantly, enabling natural language programming. Even inexperienced programmers can now replicate existing programs or build new apps from scratch.
But AI-assisted coding introduces new uncertainties—both within and between programs. “Vibe coding” abstracts away complex dependency networks in software, but this makes applications vulnerable to functional and security issues when source libraries or inputs change. Additionally, when people use AI to create personalized apps and workflows, interoperability with others’ systems becomes harder. In fact, two “vibe-coded” programs with identical functions may differ significantly in logic and output structure.
Historically, file formats and operating systems handled software consistency and compatibility, later supplemented by shared software and API integrations. But in this new era of real-time software evolution, iteration, and branching, standardization layers need broad accessibility, continuous upgradability, and sustained user trust. Furthermore, AI alone cannot solve the problem of incentivizing people to build and maintain these connections.
Blockchain addresses both issues: a protocolized synchronization layer can be embedded into users' custom software architectures and dynamically updated to maintain cross-system compatibility as environments change. Historically, large enterprises paid system integrators like Deloitte millions to customize Salesforce instances. Today, engineers can build custom sales dashboards over a weekend—but as the number of custom apps grows, developers need professional support to keep them synchronized.
(Note: Salesforce is a CRM software service provider founded in March 1999 in the United States.)
This resembles modern open-source library development, but with continuous updates (instead of periodic releases) and built-in incentives—both made easier by crypto. Like other blockchain-based protocols, shared ownership of the sync layer motivates stakeholders to invest in improvements: developers, users (and their AI agents), and other consumers can be rewarded for introducing, using, and optimizing new features and integrations.
Shared ownership also aligns all users with the protocol’s success, creating a buffer against malicious behavior—just as Microsoft wouldn’t arbitrarily break .docx standards (due to user and brand impact), co-owners of a sync layer won’t introduce inefficient or malicious code.
Like all successful software standardization architectures, this model holds massive network effect potential. As the “Cambrian explosion” of AI-coded software continues, the network of heterogeneous systems needing to communicate will grow exponentially. In short: vibe coding needs more than just a “coding style”—it needs crypto to maintain system synchronization.
7. Micropayment Systems Supporting Revenue Sharing
Author: Liz Harkavy
AI agents and tools like ChatGPT, Claude, and Copilot offer convenient new ways to navigate the digital world—but for better or worse, they're undermining the economic foundation of the open internet. We’ve seen early signs: education platforms losing traffic as students turn to AI, and U.S. newspapers suing OpenAI for alleged copyright infringement. Without rebalancing incentives, we risk a more closed internet: more paywalls, fewer content creators.
Policy solutions exist, but while they move slowly through legal channels, technical alternatives are emerging. One of the most promising (and technically challenging) approaches is embedding revenue-sharing mechanisms directly into the web’s architecture: when AI-driven actions lead to transactions, the content sources that informed those decisions should receive a share. Affiliate marketing ecosystems already do similar attribution and revenue splitting, but advanced versions could automatically trace and reward all contributors along an information chain—blockchain technology is ideally suited to track this provenance.
Such systems require new infrastructure—especially micropayment systems capable of handling multi-source microtransactions, attribution protocols that fairly assess contribution value, and governance models ensuring transparency and fairness. Many existing blockchain tools—like rollups and Layer 2s, AI-native financial institutions like Catena Labs, and financial infrastructure protocols like 0xSplits—are showing promise, enabling near-zero-cost transactions and finer-grained payment splits.
Blockchains can facilitate complex agent payment systems in several ways:
-
Nanopayments can be split among multiple data providers, allowing a single user interaction to trigger automated micro-payments to all contributing sources via smart contracts.
-
Smart contracts can enable post-execution payments triggered after a transaction, compensating information sources in a fully transparent and traceable manner.
-
Additionally, blockchains support complex, programmable payment splits, ensuring revenue is distributed fairly via code-enforced rules rather than centralized decisions—creating trustless financial relationships between autonomous agents.
As these technologies mature, they’ll unlock new economic models for media, capturing value creation across creators, platforms, and users.
8. Blockchain as an Intellectual Property and Provenance Registry
Author: Scott Duke Kominers
The rise of generative AI demands efficient, programmable IP registration and tracking mechanisms—not only to clarify ownership but also to support business models around IP access, sharing, and remixing. Existing IP frameworks rely on costly intermediaries and reactive enforcement, ill-suited for an era where AI instantly consumes content and generates variants with one click.
What we need is an open public registry—providing clear ownership proofs so creators can efficiently engage, and AI and other web applications can directly interface. Blockchain is ideal: it enables intermediary-free IP registration, provides immutable provenance records, and allows third-party apps to easily identify, license, and use protected IP.
Skepticism exists around “technology protecting IP,” given that past internet eras (and ongoing AI advances) are often associated with weakened IP protection. Partly, this stems from current IP models focusing on “preventing derivatives” rather than incentivizing and monetizing them. But programmable IP infrastructure not only lets creators, brands, and IP holders assert ownership in digital spaces—it opens doors to business models centered on IP sharing (for generative AI and other digital uses). This transforms generative AI’s main threat into an opportunity.
We’re already seeing early experiments in NFTs: companies use Ethereum-based NFTs to drive network effects and value under CC0 branding; infrastructure providers are building protocols (like Story Protocol) or even dedicated blockchains for standardized, composable IP registration and licensing. Artists are beginning to use tools like Alias, Neura, and Titles to license their styles and works for creative reuse. Incention’s Emergence series invites fans to co-create a sci-fi universe and characters, with a blockchain registry built on Story Protocol tracking creator attribution for each element.
9. Web Crawlers That Compensate Content Creators
Author: Carra Wu
Today’s most commercially relevant AI agents aren’t coding or entertainment tools—they’re web crawlers: autonomous bots that browse websites, collect data, and decide what to scrape.
Estimates suggest nearly half of today’s web traffic comes from non-human sources. Crawlers often ignore robots.txt (a file meant to guide crawler access, though it has little enforceability), using scraped data to reinforce tech giants’ market dominance. Worse, website owners bear the cost of bandwidth and CPU consumed by these uninvited visitors. In response, CDNs like Cloudflare offer blocking services—but these are makeshift fixes for a problem that shouldn’t exist.
We’ve argued that the internet’s native economic agreement between creators and distributors may be breaking down—and data confirms this trend. Over the past 12 months, website owners have increasingly blocked AI crawlers: in July 2024, only 9% of the top 10,000 sites blocked AI crawlers; today, that figure stands at 37%, and it will likely rise further as site operators gain technical means and user frustration grows.
Is there a middle ground beyond relying on CDNs to blanket-block suspected crawlers? AI crawlers shouldn’t get free access to systems built for human traffic—they should pay for data. This is where blockchain shines: each crawler agent could hold crypto and negotiate access on-chain with a site’s “gatekeeper agent” or paywall via protocols like x402 (though the challenge lies in overcoming the deeply entrenched robots.txt model from the 1990s, which may require coordinated action or CDN participation).
Meanwhile, humans could prove identity via World ID (see Chapter 3) and access content for free. In this model, content creators and site owners are compensated during AI dataset collection, while human users still enjoy the “free information” internet.
10. Privacy-Preserving Personalized Advertising
Author: Matt Gleason
AI is beginning to shape online shopping, but what if the ads we see daily were actually useful? People dislike ads for good reason: irrelevant ads are noise, while hyper-targeted AI ads based on massive consumption data feel invasive. Other models profit by locking content behind unskippable ads (like streaming services or game levels).
Crypto offers a way to rebuild advertising. AI-powered personalized agents, combined with blockchain, can strike a balance between irrelevant and overly precise ads—delivering promotions based on user-defined preferences. Crucially, this doesn’t require exposing full user data, and users can be directly compensated for sharing data or engaging with ads.
Realizing this vision requires several technical components:
-
Low-Cost Digital Payments: To compensate users for ad interactions (views, clicks, conversions), companies must send frequent micro-payments—requiring high throughput and near-zero fees;
-
Privacy-Preserving Data Verification: AI agents must prove users meet certain demographic criteria, which zero-knowledge proofs can achieve without compromising privacy;
-
Incentive Models: If the internet adopts micro-payment-based monetization (e.g., less than $0.05 per interaction, as in Chapter 7), users can opt to “watch ads for small rewards,” transforming today’s exploitative model into a participatory one.
For decades, online (and offline) advertising has chased relevance. Reimagining ads through crypto and AI may finally make them genuinely useful—personalized without being intrusive, benefiting all parties: developers and advertisers gain sustainable, incentive-aligned business models; users gain new pathways to explore the digital world.
This won’t just increase ad value—it could overturn today’s entrenched “exploitative” ad economy, building a more human-centered system where users aren’t traded commodities, but active participants.
11. AI Companions Owned and Controlled by Humans
Author: Guy Wuollet
People now spend more time on devices than in face-to-face interactions, increasingly engaging with AI models and AI-generated content. These models already offer companionship—whether for entertainment, information, niche interests, or child education. It’s easy to imagine that soon, AI companions for education, healthcare, legal advice, and emotional support will become mainstream interfaces.
Future AI companions will be infinitely patient and deeply personalized—not mere tools or robotic servants, but potentially cherished relationships. Thus, the question of “who owns and controls these relationships” becomes crucial: the user, or corporate intermediaries? If you’ve worried about content curation and censorship on social media over the past decade, this issue will soon become far more complex and personal.
The argument that “censorship-resistant, blockchain-hosted platforms are the most viable path to user-controlled AI” has been made repeatedly (as noted earlier). In theory, individuals could run device-side models or buy their own GPUs, but most lack either the resources or technical know-how.
While widespread AI companions may still be years away, the underlying technologies are advancing rapidly: text-based conversational companions are already mature, visual avatars are improving, and blockchain performance is increasing. To make censorship-resistant companions usable, we need better UX for crypto-powered apps. Fortunately, wallets like Phantom have greatly simplified blockchain interactions. Embedded wallets, password keys, and account abstraction now let users hold self-custodial wallets without memorizing seed phrases. High-throughput trusted computation via Optimistic and ZK coprocessors will also help build deeper, more enduring relationships with digital companions.
Soon, the debate won’t be about “when realistic digital companions will arrive,” but “who will control them.”
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
a16z
