AI x crypto crossovers

The economics of the internet are already changing. As the open web collapses into a prompt bar, we have to wonder: Will AI lead to an open internet or a maze of new paywalls? And who will control it — big centralized companies or broad communities of users?

That’s where crypto comes in. We’ve discussed the intersection of AI and crypto quite a bit; but briefly, blockchains are a new way to architect internet services and build networks that are decentralized, credibly neutral, and ownable by users. They offer a counterbalance to many of the centralizing forces we’re already seeing among AI systems by renegotiating the economics that underpin today’s systems, helping achieve a more open, more robust internet. 

The idea that crypto can help build better AI systems, and vice versa, isn’t new — but it’s often been poorly defined. Some intersecting areas — like verifying “proof of humanity” given the proliferation of low-cost AI systems — are already attracting builders and users. But other use cases seem years, if not decades, away. So in this post, we share 11 use cases at the intersection of crypto and AI to help kickstart conversations about what’s possible, what challenges are left to solve, and more. They’re all grounded in technology being built today, from processing a multitude of micropayments to ensuring humans own their relationships with the AI of the future.

CATEGORIES

IdentityDecentralized Infrastructure for AINew Economic and Incentive ModelsOwning future AI

IDENTITY

by Scott Duke Kominers

Generative AI thrives on data, but for many applications, context — the state and background information relevant to an interaction — is equally, if not more, important.

Ideally, an AI system — whether agent, LLM interface, or other application — would remember the kinds of projects you’re working on, your communication style, and your preferred programming languages, among a multitude of other details. But in practice, users often have to reestablish this context across different interactions within a single application — like when you launch a new ChatGPT or Claude shell — let alone when moving between systems. 

At the moment, context from one generative AI application is rarely, if ever, portable to others. 

With blockchains, AI systems could enable key context elements to exist as persistent digital assets, which can be loaded in at the start of a session and transferred seamlessly across AI platforms. Moreover, blockchains are potentially the only solution to this problem that is both forwards-compatible and that establishes a commitment to interoperability, given these features are defining properties of blockchain-based protocols. 

A natural application for this is AI-mediated gaming and media, where preferences (from difficulty levels to key bindings) could persist across different games and environments. But the real value is in knowledge applications, where AIs need to understand what users know and how they learn; as well as in more professionalized AI use cases, like coding. Of course individual businesses already develop their own customized bots with global context specific to a given business — but in that case, the context generally isn’t portable, even across the different AI systems used within the organization. 

Organizations are only just starting to understand this issue, and the closest general-purpose solutions we’ve seen so far are custom bots with fixed, persistent context. But portability of context between users within-platform is starting to emerge offchain; with Poe, for instance, users can rent out their custom bots to others. 

Bringing this sort of activity onchain would make it possible for the AI systems we interact with to share a context layer comprising the key elements from all of our digital activity. They would understand our preferences immediately, and be better able to fine-tune and optimize our experience. And conversely, as with onchain intellectual property registries, enabling AI to reference persistent onchain context creates the possibility of new and better marketplace interactions around prompts and information modules — for example, users could license or monetize their expertise directly, while retaining custody of their data. And of course, shared context will make many things possible that we haven’t even conceived of yet. 

by Sam Broner

Identity, the canonical record of who or what a thing is, is the silent plumbing that enables today’s digital discovery, aggregation, and payment systems. Because platforms keep this plumbing behind walls, we experience identity as part of a finished product: Amazon assigns identifiers (ASIN or FNSKU) to products, lists products in one place, and helps users discover and pay. Facebook is similar: A user’s identity is the basis for their feed and for discovery across the app, including Facebook Marketplace listings, organic posts, and paid advertising.

This is all poised to change as AI agents advance. As more companies use agents — for customer service, logistics, payments, and other use cases — the less their platforms will look like single-surface apps. Instead they’ll live across multiple canvases and platforms, accumulate deep context, and perform more tasks for users. But tying an agent’s identity to only one marketplace makes it unusable everywhere else that matters: email threads, Slack channels, and within other products.

That’s why agents need a single, portable “passport”. Without one, there’s no way to know how to pay the agent, verify its version, query its capabilities, know who the agent is working on behalf of, or trace its reputation across apps and platforms. An agent’s identity needs to act as wallet, API registry, changelog, and social proof — so any interface (email, Slack, another agent) can resolve and speak to it the same way. Without the shared primitive of “identity”, every integration needs to rebuild this plumbing from scratch, discovery remains ad-hoc, and users lose context each time they switch channels or platforms.

We have the opportunity to design agentic infrastructure from first principles. So how do we build a credibly neutral identity layer that is richer than a DNS record? Instead of reinventing monolithic platforms — where identity is combined with discovery, aggregation, and payment — agents should be able to accept payments, list capabilities, and exist in multiple ecosystems without fear of being locked in to any particular platform. This is where the intersection of crypto and AI is especially useful, because blockchain networks provide permissionless composability, which can allow builders to create more useful agents and better user experiences.

In general, vertically integrated solutions, like Facebook or Amazon, currently have better UX — part of the inherent complexity of building a great product is making sure the pieces make sense together, top down. But the price of that convenience is high, especially as the cost of building the software to aggregate, market, monetize, and distribute agents falls and the surface area for agentic apps expands. It’ll take work to match the UX of vertically integrated providers, but a credibly neutral identity layer for agents would allow entrepreneurs to own their own passport — and encourage experimentation in distribution and design.

by Jay Drain Jr. and Scott Duke Kominers

As AI becomes more pervasive — powering bots and agents across all manner of web interactions, including deepfakes and social media manipulation — it’s increasingly difficult to know whether you’re interacting with a real human online. This erosion of trust isn’t a future concern; it’s already here. From X feed comment armies to bots on dating apps, reality is starting to blur. In this environment, proof of personhood becomes essential infrastructure.

One way to prove you’re human is through digital IDs (including the centralized ones used by the TSA). Digital IDs encompass all of the things a person can use to verify their identity — usernames, PINs, passwords, and third-party attestations (e.g., citizenship or creditworthiness) and other credentials. The value of decentralization here is clear: When this data lives in centralized systems, issuers can revoke access, impose fees, or abet surveillance. Decentralization flips this dynamic: Users, not platform gatekeepers, control their own identities, making them more secure and censorship-resistant.

Unlike traditional identity systems, decentralized proof of personhood mechanisms (like World’s Proof of Human) allow users to control and custody their own identities, and verify their humanity in a privacy-preserving and credibly neutral way. And like a driver’s license, which can be used anywhere regardless of when or where it was issued, decentralized PoP can serve as a reusable base layer across any platform, including ones that don’t exist yet. In other words, blockchain-based PoP is forwards compatible because it offers:

1. Portability: Protocols are public standards that any platform can integrate. Decentralized PoP can be managed via public infrastructure, and is within the user’s control. This makes it completely portable, and any platform can be made compatible with it now or in the future.
2. Permissionless accessibility: Platforms can independently choose to recognize the PoP ID, without having to pass through a gatekeeper API that might discriminate against different use cases.

The challenge in this space is adoption: While we haven’t yet seen many real-world proof of personhood use cases with meaningful scale, we anticipate a critical mass of users, handful of early partnerships, and killer apps will accelerate adoption. Every application that leverages a given digital ID standard makes that ID type more valuable to users; this drives more users to obtain the ID; which in turn makes the ID more attractive for applications to integrate as a way to certify personhood. (And because onchain IDs are interoperable by design, these network effects can grow quickly.)

We’ve already seen mainstream consumer applications and services in gaming, dating, and social media announce partnerships with World ID to help humans know that they’re playing, chatting, and transacting with real humans — indeed, the particular humans they’re expecting. We’ve also seen new identity protocols emerge this year, including the Solana Attestation Service (SAS). Though not a proof of personhood issuer, SAS allows users to privately associate off-chain data — like KYC checks for compliance or accreditation status for investing — with Solana wallets in order to build users’ decentralized identity. All of this suggests the inflection point for decentralized PoP may not be far off. 

Proof of personhood isn’t just about banning bots, it’s about establishing clear boundaries between AI agents and networks of humans. It enables users and applications to distinguish between human and machine interactions, creating space for better, safer, and more authentic digital experiences.

DECENTRALIZED INFRASTRUCTURE FOR AI

by Guy Wuollet

AI may be a digital service, but its advancement is increasingly bottlenecked by physical infrastructure. Decentralized Physical Infrastructure Networks, or DePIN — which offers a new model for building and operating real-world systems — can help democratize access to the compute infrastructure underlying AI innovation, making it cheaper, more resilient, and more resistant to censorship. 

How? Two of the biggest barriers to AI advancement have been energy and access to chips. Decentralized energy can help make more power available, but builders are also using DePIN to aggregate unused chips from gaming PCs, data centers, and other sources. These computers can come together to form a permissionless compute marketplace, leveling the playing field for building new AI products. 

Other use cases include distributed training and fine tuning of LLMs, as well as distributed networks for model inference. Decentralized training and inference can potentially lead to much lower costs because they use otherwise latent compute. They can also provide censorship resistance, ensuring that developers don’t get de-platformed by hyperscalars — large-scale, centralized cloud service providers that offer massively scalable computing infrastructure.

Centralization of AI models among a handful of companies is a persistent concern; decentralized networks can help create more cost effective, more censorship resistant, and more scalable AI. 

by Scott Duke Kominers

As AI tools become better at solving complex tasks and carrying out multi-layered chains of interaction, AIs will increasingly need to interact with other AIs, independent of human controllers. 

For example, an AI agent may need to request specific data relevant to a computation, or recruit specialized AI agents for particular tasks — such as assigning a statistics-bot to develop and run model simulations, or engaging an image-generation bot in the process of creating marketing materials. AI agents will also create significant value in completing the entire flow of a transaction or any other activity on behalf of a user — like finding and booking a plane ticket according to someone’s preferences, or discovering and ordering a new book from their favorite genre. 

Today there are no established, generalized agent-to-agent markets — these sorts of cross-queries are mostly only available via explicit API connections, or within AI agent ecosystems that maintain agent-to-agent calls as an internal functionality. 

More broadly, most of today’s AI agents operate in siloed ecosystems, with somewhat closed APIs and a general lack of architecture standardization. But blockchain technologies can help protocols establish open standards, which is important for adoption in the short-run. Longer-term, this also supports forwards compatibility: As new types of AI agents evolve and are created, they can expect to be able to plug into the same underlying network. Blockchains can more readily adapt across novel AI innovations, given their interoperable, open source, decentralized, and often more easily upgradeable architectures.

A number of companies are already building blockchain rails for agent-to-agent interactions as the market develops: Halliday, for example, recently introduced its protocol providing standardized, cross-chain architecture for AI workflows and interactions — with protocol-level protections to ensure that the AI doesn’t go beyond the user’s intent. Catena, Skyfire, and Nevermind, meanwhile, use blockchains to support payments from one AI agent to another without humans needing to be in the loop. Many more such systems are in progress, and Coinbase has even started providing infrastructure support for these efforts.

by Sam Broner and Scott Duke Kominers

The recent revolution in generative AI has made it easier than ever to build software. Coding is orders of magnitude faster, and — perhaps most importantly — can be done in natural language, so that even inexperienced programmers can fork existing programs and build new ones from scratch. 

But while AI-assisted coding creates these novel opportunities, it also introduces a lot of entropy both within and across programs. “Vibe coding” abstracts away the complex web of dependencies underlying software — but this can also leave programs vulnerable to both functionality and security deficiencies as source libraries and other inputs change. Meanwhile, when people use AI to create their own individually customized applications and workflows, it becomes harder for them to interface with other people’s systems. Indeed, even two vibe-coded programs that in effect do the same task may have very different operations and output structures.

Historically, standardization to ensure consistency and compatibility was provided first by file formats and operating systems, and more recently by shared software and API integrations. But in a world where software is evolving, morphing, and branching in real time, the standardization layers will need to be broadly accessible and constantly upgradable — all while maintaining user trust. Moreover, AI alone does not solve the problem of incentivizing people to build and maintain these linkages.

Blockchains provide an answer to both problems at once: protocolized synchrony layers, which are wrapped into people’s custom software builds and update dynamically to ensure cross-compatibility as things change. Historically, a large enterprise might pay a “system integrator” like Deloitte millions to customize a Salesforce instance. 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 operational.

This is similar to the way that open source software library development works today, except with ongoing updates rather than periodic releases — and an incentive wrapper. Both of which are made more easily possible with crypto. Just like with other blockchain-based protocols, shared ownership of synchrony layers incentivizes active investment in improving them. Developers, users (and/or their AI agents), and other consumers can be rewarded for introducing, using, and evolving new features and integrations. 

And conversely, shared ownership gives all users a stake in the overall success of the protocol, which serves as a buffer against malbehavior. Just like Microsoft is disincentivized from corrupting the .docx file standard because of the ripple effects on its users and brand, co-owners of a synchrony layer are disincentivized from introducing clumsy or malicious code into the protocol. 

As with all the software standardization architectures we’ve seen previously, there’s tremendous potential for network effects here. As the Cambrian explosion of AI-coded software continues, the web of heterogeneous, diverse systems that need to stay in communication with each other will expand dramatically. In short: vibe coding needs more than just vibes to stay in sync. Crypto is the answer. 

NEW ECONOMIC AND INCENTIVE MODELS

by Liz Harkavy

AI agents and tools like ChatGPT, Claude, and Copilot promise a convenient new way to navigate the digital world. But, for better or worse, they are also destabilizing the economics of the open internet. We’re already seeing this play out — for example, educational platforms are seeing significant traffic declines as students increasingly use AI tools, and several U.S. newspapers are suing OpenAI for copyright infringement. If we don’t realign incentives, we could see an increasingly closed internet, with more paywalls and fewer content creators. 

There are always policy solutions, of course, but while those work their way through the courts, a number of technical solutions are presenting themselves. Perhaps the most promising (and technically complex) solution is to build a system of revenue sharing into the architecture of the web. When an AI-driven action leads to a sale, the content sources that informed that decision should receive a cut. The affiliate marketing ecosystem already does attribution tracking and revenue-sharing like this; a more sophisticated version could automatically track and reward all contributors in the chain of information. Blockchains can obviously play a role in tracking that chain of provenance.

But a system like this requires new infrastructure with other features as well — in particular, micropayment systems capable of handling tiny transactions across many sources, attribution protocols that fairly value different kinds of contributions, and governance models that ensure transparency and equity. Many existing blockchain-based tools — such as rollups and L2s, AI-native financial institution Catena Labs, and financial infrastructure protocol 0xSplits — show potential here, enabling near-zero-cost transactions and more-finely-grained payment splits. 

Blockchains would enable sophisticated agentic payment systems through several mechanisms:

• Nanopayments can be split across multiple data providers, allowing a single user interaction to trigger tiny payments to all contributing sources through automated smart contracts. 
• Smart contracts allow for enforceable retroactive payments triggered by completed transactions, compensating information sources that contributed to a purchase decision after the transaction occurs with full transparency and traceability. 
• Additionally, blockchains enable distribution of complex and programmable payment splits, ensuring that revenue is fairly allocated through code-enforced rules rather than centralized decisions, creating trustless financial relationships between autonomous agents.

As these emerging technologies mature, they can create a new economic model for media that captures the full chain of value creation, from creators to platforms to users. 

by Scott Duke Kominers

Generative AI has created an urgent need for efficient and programmable mechanisms for registering and tracking intellectual property — both to be certain of provenance and to enable business models around IP access, sharing, and remixing. Existing IP frameworks — which rely on costly intermediaries and ex-post enforcement — are undermatched for a world where AI consumes content instantly and generates new variations with just a click.

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 interface with directly. Blockchains are ideal for this because they make it possible to register IP without relying on an intermediary, and provide immutable proof of provenance; they also make it straightforward for third-party applications to recognize, license, and interact with that IP. 

There’s understandably a lot of skepticism around the entire idea that technology can somehow protect IP when the first two eras of the web — as well as the ongoing AI revolution — have often been associated with decreases in intellectual property protection. One issue is that many of today’s IP-based business models have been focused on excluding derivative works, rather than trying to incentivize and monetize them. But programmable IP infrastructure doesn’t just make it possible for creators, franchises, and brands to clearly establish ownership of their IP in digital space — it also opens the door to business models explicitly centered around sharing IP for use in generative AI and other digital applications. In effect, this turns one of generative AI’s primary threats to creative work into an opportunity.

We’ve already seen creators experiment with newer models early-on in the NFT space, with companies leveraging NFT assets on Ethereum to support network effects and value accrual under CC0 brand-building. More recently, we’ve seen infrastructure providers building protocols and even specialized blockchains (e.g., Story Protocol) for standardized and composable IP registration and licensing. Some artists have already started using these tools to license their styles and work for creative remixing via protocols like Alias, Neura, and Titles. Incention‘s Emergence franchise, meanwhile, engages its fan base in co-creating a sci-fi universe and its characters, with a blockchain registry built on Story that keeps track of who created what.

by Carra Wu

Today, the AI agent with the best product-market fit isn’t an agent for coding or for entertainment. It’s the webcrawler — autonomously navigating the web, collecting data, and making decisions about which links to follow.

By some estimates, nearly half of all internet traffic now originates from non-human sources. Bots routinely flout the niceties of robots.txt — a file that is supposed to inform automated web crawlers whether they are welcome on a site, but in practice has very little authority — and use the data they extract to power the defensibility of some of the biggest tech companies on the planet. Worse still, websites end up footing the bill for these uninvited guests, paying to serve bandwidth and CPU resources to what can feel like an endless tide of faceless scrapers. In response, companies like Cloudflare and other CDNs (content delivery networks) provide blocking services. It’s a patchwork of services that shouldn’t need to exist. 

We’ve argued before that the original agreement of the internet — the economic covenant between the creators who make content and the platforms that distribute it — is likely to come undone. This is starting to show up in the data: Over the past twelve months, website owners have started blocking AI-oriented scrapers in droves. Where, in July of 2024, only around nine percent of the top 10,000 websites barred AI crawlers, that figure is now at 37% percent. It will only increase as more website operators get sophisticated and users continue to get frustrated. 

So what if, instead of paying CDNs to outright block anyone who seems like they might be a bot, we meet somewhere in the middle? Instead of freeloading off of a system meant to drive human traffic to sites, AI bots could pay for the right to collect data. This is where blockchains come in: In this scenario, every webcrawler agent would have some crypto, and engage in an onchain negotiation with the “bouncer” agent or paywall protocol of each website via x402. (The challenge, of course, is that the robots.txt system, also known as the Robots Exclusion Standard, has been entrenched in the way internet companies have done business since the 1990s. It would take large scale group coordination, or participation on the part of a CDN like Cloudflare to overcome this).

But humans, in a separate lane, could prove their humanity through World ID (see above) and gain access to content for free. In this way, content creators and website owners could be compensated for their contributions to large AI datasets at the point of collection, and humans could continue to enjoy an internet where information wants to be free.

by Matt Gleason

AI has already started to impact how we shop online, but what if the ads we see every day were… helpful? People don’t like ads for a number of obvious reasons. Offbase ads are pure noise. At the same time, not all personalization is created equal. AI-powered ads that are too targeted — drawing from troves of consumer data — can feel invasive. Other apps try to monetize by gating content (streaming content services or game levels, for example) behind unskippable ads.

Crypto can help address some of these problems, offering the opportunity to reimagine how advertising works. Paired with blockchains, personalized AI agents can close the distance between irrelevant and uncanny, delivering ads based on user-defined preferences. But importantly, they can do this without globally exposing user data and while compensating users who share data or engage with ads directly.

Some technological requirements here include:

1. Low-fee digital payments: To compensate users for ad interactions (watching, clicking, converting), companies need to send small, frequent payments. For this to work at scale, we need fast, high-throughput systems with negligible fees.
2. Privacy-preserving data verification: AI agents need to be able to prove that consumers meet some demographic attributes. Zero-knowledge proofs can verify demographic attributes while preserving privacy. 
3. Incentive models: If the internet embraces micropayment-based monetization (e.g., <$0.05 per interaction, described above), users would be able to opt into ads in exchange for small payments, flipping the current model from extraction to participation.

People have been trying to make ads relevant for decades online — and for centuries offline. But rethinking ads through the lens of crypto and AI can finally make ads more useful. Tailored without being creepy, and in a way that benefits everyone: For builders and advertisers, it unlocks new incentive structures that are more sustainable and aligned. And for users, it provides more ways to discover and navigate their digital worlds.

All of this would make ad space more valuable, not less. It could also unseat today’s deeply entrenched, extractive ad economy, and replace it with something more human-centric: a system where users are treated as participants, not products.

OWNING THE FUTURE OF AI

by Guy Wuollet

Many people spend more time on their devices than interacting in-person, and this time is increasingly spent interacting with AI models and AI-curated content, in particular. All of these models are already providing a form of companionship, whether that’s entertaining, informing, indulging a niche interest, or teaching kids. It’s easy to imagine a near-future where AI-based companions for education, healthcare, legal counsel, and friendship become a popular mode of interaction for humans. 

AI companions of the future would be infinitely patient, and tailored to a specific individual and their specific use case. Beyond being just helpers or robot servants, they could become highly valued relationships. So the question of who will have ownership and control over these relationships — whether users or companies and other intermediaries — becomes equally important. If you were already concerned about social media curation and censorship in the past decade, the problem will get exponentially more complicated, and more personal, in the future. 

It’s not a new argument (already described here and here) that censorship-resistant hosting platforms like blockchains offer the most cogent path towards uncensorable, user-controlled AI. It’s true that individuals could run on-device models and buy their own GPUs, but most people either can’t afford to, or simply don’t know how. 

Although we’re still a ways out from widely prevalent AI companions, the technologies for all this are improving quickly: Seemingly human, text-based companions are already excellent. Visual avatars have improved significantly. Blockchains are becoming more performant. To ensure that uncensorable companions are easy to use, we’ll need to rely on better UX for crypto-powered applications. Thankfully wallets (like Phantom) have made interacting with a blockchain much simpler, and embedded wallets, passkeys, and account abstraction make it possible for users to hold self custodied wallets without the complexity of storing a seed phrase themselves. Technologies like high-throughput trustless computers, using optimistic and ZK coprocessors, will also make it possible to build meaningful and lasting relationships with digital companions. 

In the near future, expect the conversation to shift from when we’ll see virtually lifelike digital companions and avatars to who and what will be able to control them. 

Be sure to subscribe to our newsletter for more trend updates, builder guides, industry reports, and other resources.

CONTRIBUTORS

Sam Broner is a partner on the investing team at a16z crypto. Prior to joining a16z, Sam was a software engineer at Microsoft, where he was on the founding team of the Fluid Framework and Microsoft Copilot Pages. Sam also attended MIT’s Sloan School of Management, where he worked on Project Hamilton at the Federal Reserve Bank of Boston, led the Sloan Blockchain Club, directed Sloan’s first AI Summit, and won MIT’s Patrick J. McGovern award for creating an entrepreneurial community.

Jay Drain Jr. is an investment partner at a16z crypto, where he focuses on consumer and application-layer projects. Prior to joining a16z in 2022, Jay was an investor at Maven Ventures, a seed-stage consumer software fund, where he led the firm’s web3 investing. Previously, Jay worked in Goldman Sachs’ Global Markets Division for two years. He graduated from Amherst College, where he studied Political Science & Law, and was a member of the Men’s Track & Field team.

Matt Gleason is a security engineer for a16z crypto, helping portfolio companies with their application security, incident response, and other audit or security needs. He has conducted audits, and found and helped fix critical vulnerabilities in code prior to project deployment on many different projects.

Liz Harkavy is a partner on the a16z crypto investment team, focusing on picks and shovels projects, as well as decentralized infrastructure. Prior to joining a16z crypto, Liz was a founding engineer at Corsali (now Vana), a venture-backed full-stack machine learning platform that is powered by highly educated workers labeling data to earn crypto from their phones. Prior to that she spent time at Facebook and JPL. Liz completed her bachelor’s in Physics and Computer Science, as well as her master’s in Computer Science, at MIT.

Scott Duke Kominers is the Sarofim-Rock Professor of Business Administration at Harvard Business School, a Faculty Affiliate of the Harvard Department of Economics, and a Research Partner at a16z crypto. He also advises a number of companies on web3 strategy, as well as marketplace and incentive design; for further disclosures, see his website. Kominers’s first book — The Everything Token: How NFTs and Web3 Will Transform the Way We Buy, Sell, and Create, co-authored with Steve Kaczynski — is now available.

Carra Wu is a partner on the a16z crypto investment team. Carra focuses on gaming, metaverse, media, and DAO investments across crypto. Previously, she spent time as a software engineer building AR/VR apps and games for the Hololens at Microsoft — and as a product manager working on search systems and developer facing products for the App Store. Carra studied Applied Math, Computer Science & Economics at Harvard University, where she danced with the Harvard Ballet Company.

Guy Wuollet is a partner on the a16z crypto investment team. He focuses on investing across crypto at all layers of the stack. Prior to joining a16z, Guy worked on independent research in concert with Protocol Labs. His work focused on building decentralized networking protocols and upgrading internet infrastructure. He holds a B.S. in Computer Science from Stanford University, where he rowed on the Varsity Crew team.

***
The views expressed here are those of the individual AH Capital Management, L.L.C. (“a16z”) personnel quoted and are not the views of a16z or its affiliates. Certain information contained in here has been obtained from third-party sources, including from portfolio companies of funds managed by a16z. While taken from sources believed to be reliable, a16z has not independently verified such information and makes no representations about the current or enduring accuracy of the information or its appropriateness for a given situation. In addition, this content may include third-party advertisements; a16z has not reviewed such advertisements and does not endorse any advertising content contained therein. 

This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. You should consult your own advisers as to those matters. References to any securities or digital assets are for illustrative purposes only, and do not constitute an investment recommendation or offer to provide investment advisory services. Furthermore, this content is not directed at nor intended for use by any investors or prospective investors, and may not under any circumstances be relied upon when making a decision to invest in any fund managed by a16z. (An offering to invest in an a16z fund will be made only by the private placement memorandum, subscription agreement, and other relevant documentation of any such fund and should be read in their entirety.) Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by a16z, and there can be no assurance that the investments will be profitable or that other investments made in the future will have similar characteristics or results. A list of investments made by funds managed by Andreessen Horowitz (excluding investments for which the issuer has not provided permission for a16z to disclose publicly as well as unannounced investments in publicly traded digital assets) is available at https://a16z.com/investments/.

The content speaks only as of the date indicated. Any projections, estimates, forecasts, targets, prospects, and/or opinions expressed in these materials are subject to change without notice and may differ or be contrary to opinions expressed by others. Please see https://a16z.com/disclosures for additional important information.