We are in the middle of two significant technological shifts happening simultaneously and also intertwined. AI and web3 may have different starting points and timelines but they are both fundamental advances in how computers can transform what we are able to do and how we can do it. They also have both shifted from unique verticals to horizontal ingredients. At USV, we’ve been investing in web3 and AI for many years, but recently we moved away from categorizing our portfolio in these buckets. Instead, we think of them as core components or foundational infrastructure, often necessary to best accomplish a utility for a customer. You can no longer transfer money across borders as quickly or cheaply on web2 architecture as you can utilizing stablecoins. You can’t record and process notes from a meeting or a physician’s interaction with a patient as efficiently as you can with AI. There are long lists of applications of each we are interested in but, maybe most of all, we are excited that these two tectonic shifts are on a collision course and set to reinforce each other. In a moment where two huge and relatively new atomic units of creation are charging forward at the same time, the moment where they collide will create a tremendous amount of new energy.
Many of us at USV have been thinking and writing about the opportunity to reinvent consumer experiences, products, and networks, both onchain and online. Onchain, platforms like Farcaster allow users to own their identities, transport audiences, and integrate economics. Simultaneously, AI is opening up the floodgates of a new age of consumer social–reintroducing the possibility of applications that start as weird, unusual, fun, quirky, and over time become fixtures of how we live and communicate. We’ve been busy dreaming about and, slowly, investing in both. But, increasingly, these two worlds are going to converge.
While there will no doubt be interesting AI applications delivered by incumbents who can roam freely across all of our data trapped in their monolithic structures, some of the most interesting AI-driven consumer experiences will rely on decentralized infrastructure that enables us to be stewards of our own data, control how we want to use and customize the agents to our liking, and permit many new applications to repeatedly dip into the same personal data wells. Imagine a travel AI-agent perfecting an itinerary because it’s pulled my food preferences from Blackbird, conversations from Farcaster channels, and photo library stored onchain. Or a personalized health service that is able to immediately understand all my biometrics from across wearables. Or an AI therapist that isn’t just cheaper but begins with the context on my schedule, conversations, and daily life that I want it to innately understand. In each, if built on web3 architecture, integrated wallets allow each experience to seamlessly become transactional, creating new opportunities on platforms.
The big dream of consumer AI-driven applications and agents are products and services that feel tailor-made. The distribution is universal but the experience is unique, crafted by each user’s own relevant data. Ideally, these products get better and better over time, as their own models improve and the data they can pull from expands. It can happen slowly, by using a product enough for it to learn our preferences and behavior. Or it can happen quickly, by leveraging data that already exists and transforming it to the use case. To do that, however, we have to own, or at least control, our data sets. Picture a world where when a new awesome and weird consumer application comes along we could simply authorize it to dip into our data and pull out what’s relevant, immediately delivering a personalized product, no new training required. It is unlikely that we will be able to do that with data locked inside platforms where the fundamental business depends on data control but very likely we will be able to on top of platforms like Farcaster or others built with this explicit ownership intent.
We are excited by all that agents are set to do for us. We’re even more excited by agents leveraging data we can authorize, reuse, and build on to supercharge what they can do, how well they can do it, and how personalized it can feel. If you’re building at this intersection, we’d love to meet you.
