Notes on Consumer Crypto | Nov 15, 2024

Welcome to my weekly consumer crypto braindump of things I'm thinking about and have bumped into during my internet travels.

Here's what I found worth talking about this week...

Jeffy Yu, creator of Zerebro, wrote a great post this week that dives into the idea that the dawn of web4 is here, one where the internet will anticipate, plan, and act autonomously. It’s building upon the social foundations of web2 and the financial foundations of web3 to unlock a world where artificial intelligence and agentic actors are ubiquitous. He goes into a bunch more detail in the post about what’s possible today, where it’s still falling short, and what it’s going to take to reach AGI, but his core point is that we’re actually further along the path than most people realize.

I know I know you’re sick of all of this agent talk, but I don’t care. I’m completely captivated. 

The two most interesting things to me in the post are his perspective on where we are on the 5 steps towards AGI that OpenAI outlined and also how crypto has quickly become a central playground for AI experimentation.

On the first, though OpenAI claims we’re in Level 2 (see below), anyone spending time in crypto these days knows we’ve firmly entered Level 3. Obviously the agents we’re seeing today leave a lot to be desired as they still often require some human oversight and are only autonomous in pretty narrowly defined use cases. But the fact that we have agents like Zerebro and Cents that are better tweeters and reply guys than 99% of users, are launching their own tokens, and creating and minting their own art is crazy. These are autonomous agents in every sense of the word.

Projects like Fungi and Ai16z (more on them in a minute) are also very close to delivering agents with true financial autonomy that will trade, learn, and adjust strategies on the fly. They’ll converse with people, build influence on twitter and warpcast, accrue financial resources, and pay people to perform tasks they can’t. 

We’re also seeing signs of Level 4 capabilities in creative areas like Truth Terminal creating its own religion, Zerebro creating its own PFP collection, and Botto continuously refining its style over years of art production. Obviously being able to innovate in more broadly defined contexts like science and technology will be the big breakthrough here.

Jeff likens AGI to the invention of electricity. It took way more than just the invention of raw electric power (the foundational models) for it to actually become wildly useful in everyday life. It needed the lightbulb, the motor, the grid, etc for it to actually be revolutionary and change the fabric of society.

And this is why crypto is going to play such a critical role in its evolution. Permissionless and standardized financial and application rails plus chaotic capital formation capabilities plus a culture that gets off on the weirdest possible shit is an incredible playground for experimentating with novel applications of super intelligence.

We’ve officially entered an agent-infrastructure cycle where foundational model improvements are unlocking new agent capabilities which inevitably hit a wall but inspire more infrastructure development which then pushes capabilities further, etc, etc.

And once agents are running amok in DeFi, running effective token trading strategies, creating wildly adopted memecoins and NFTs, and managing DAOs the trad world will start bending over backwards to figure out how to bring them over. Them bringing their products onchain is the exciting answer for all of us, but agents are obviously also going to get better at interfacing with traditional systems and the real world so we’ll see.

Then ultimately the massive disruptive shift will happen when traditional corporations aren’t just improving efficiencies by leveraging AI, but actually need to start serving Agentic Protocols to survive as they get increasingly important to the economic machine.

So the important takeaway here is to take all this weird experimentation we’re seeing seriously. It’s easy to poke holes in it today, but those holes are gonna get filled. The question to focus on is what happens when all of this starts really working.

ai16z is a fully AI-managed tokenized venture DAO where agents evaluate investment opportunities suggested to it by the community, execute trades, evaluate the reputations of members based on the success of those trades, and grow the AUM of the fund. It’s been on a crazy speculative run over the past few weeks on the backs of a big vision and strong memetics (ai16z, marc aindreessen, flip a16z). They’re expected to have the agents ready to start trading and their “virtual trust marketplace” live in the next few weeks.

I find Ai16z so fascinating because it’s so much more than just the fund it appears to be on the surface. It’s an attention network and agentic schelling point. It’s a memecoin, an agent launchpad, and social network. And the flywheels that are going to kick off when it starts trading are going to melt faces.

Let me explain.

Tech Flywheel. They’ve built an open source framework for launching agents, called Eliza, that many of the top sentient coins are building on top of and contributing to. It’s becoming a no brainer tech stack for people wanting to build crypto agents which is quickly becoming everybody. They also have a “standard deal” where if you build on Eliza and donate 10% of the token supply to the DAO you’ll get added to their portco list and be supported by the project. 

So as more people want to tap into the attention base of the project, more people will build on the tech, donate tokens, and grow the AUM of the DAO. As more people build on the tech, the tech will improve and be extended in new ways, which will make it increasingly attractive to build on top of and further grow the attention base of the project. Larger attention base, more builders, more donations, better tech, more attention, etc.

Social Flywheel. Their virtual trust marketplace allows token holders to suggest trades to the agents, have their reputation evaluated as a source of insights, and grow their influence as their suggestions yield good results. Not only will having a good reputation score and ranking high on the leaderboard be socially desirable, but being able to influence the agent to buy things you own will be very economically desirable. 

So I expect many people to want to play this game which drives demand for the token. People will shill their rankings, which will attract more people to the game, which will create a richer social experience, which will reduce sell pressure and further drive more demand for the token.

Economic Flywheel. The core purpose of the DAO is to become a very effective trading firm and grow AUM. All assets the DAO is holding are publicly viewable and with the levels of attention the project has any trade it makes will rapidly be copied (just as is the case with any trader with good reputation and a tg channel where they share their calls). 

So the virtual trust marketplace will make sure the agents see everything, they’ll get increasingly good at choosing the right trades, those trades will get copied guaranteeing memes break out and kicking off market reflexivity, AUM will grow rapidly, attention will follow, more people will copy trade, etc.

Obviously there’s still a lot to figure out here (reputation systems are hard, how do you exit trades elegantly, etc), but the potential for this to get insane in a hurry are there. It’s trading at over 100x its current AUM which sounds crazy, but is it?

If you’ve been paying attention to us or are active in the Ethereum ecosystem you probably know Interface. It’s a social trading product based on a feed of onchain activity (with a for you and following feed like twitter). They’ve been building for years and have had a good group of core users, but they’ve hit an inflection point and have been consistently growing 50% WoW for the past few months. 

Interface came through SC06 in late 2023. They were a strong technical team of crypto natives with a unique insight into what onchain social was going to look like. I’ve long held the belief that onchain transactions are fundamentally social and are a better starting point for a new social network than competing on content. 

In the year since they’ve continued to make great product progress and it’s gotten increasingly useful for people that spend a ton of time onchain, but they hadn’t figured out how to breakthrough into broader audiences. They didn’t have a killer hook and a fast path to an “aha” moment that reinforces it. 

But a couple of months ago they introduced a “copy trade” feature where you could seamlessly copy trade any swap directly from the feed and trading fees are earned by the originator of the trade. Since introducing this their growth and daily active usage has hit an inflection point.

Now when talking about Interface you can say “it’s the easiest place to find and trade alpha”. You can download the app, follow suggested people based on their pnl, instantly see new trades, and copy trade and make money. 

After this experience has solidified the value of the product you can start to find your friends, build up a following, use it to explore the onchain world, and have more socially-focused experiences. 

But the power of a really strong hook that taps into a deep desire (make money, fomo) and a product designed to close the loop on that experience as quickly as possible can’t be overstated.

Hyperliquid is an L1 and decentralized perps exchange with “best-in-class speed, liquidity, and price”. I’ve been using it for a while as I wanted to get my hands dirty and understand the perps game, and after trying out a handful of them it was just so clearly the best. Their volume and TVL have grown substantially over the past year, and from what I can see looking at DefiLlama they do the more derivatives volume than any other chain. They’re imminently launching their HYPE token and have some unique properties that make it something you need to pay attention to.

We all know the story of the low float high fdv chain launches that have dominated the past few years. Raise a ton of private money, run farming campaigns on testnet, airdrop a small percentage of tokens to farmers alongside CEX listings at huge FDVs, investors get rich and your community gets shackles at best or rekt at worst, nobody builds anything useful on the chain.

Hyperliquid is doing something completely different:

• They’ve raised no VC money at all

• They built a truly killer best-in-class product in a huge category alongside building the L1

• The dev team has taken no fees and it’s all accruing to the protocol

• They’re launching their token with a high float with a large percentage going to early users

It would have been so easy for them to raise a ton of money and had a fat payday already, but they’ve completely rejected all these short term incentives to set themselves up with the best chance to succeed long term. They’ve recognized how important it is to maintain neutrality and not have insiders on the path to becoming the “platform where all finance happens”. 

I really hope this succeeds and it can become a powerful blueprint for other teams to do the same.

Techcrunch shared a good article this week highlighting some high level trends and app adoption metrics among Gen Z in 2024. 

Anyone building in consumer, especially if you’re building in consumer social, needs to have a great pulse on what Gen Z cares about and is using. They’re the trend setters and are in a growth stage of their life where they’re socially active, constantly forming new connections, and very open to trying new things. 

Nikita Bier famously shared the stat that share rates drop 20% for every additional year of age from 13 - 18, so if you’re targeting anyone much older than that you need to expect to pay for acquisition. All breakout social products in recent history have gone viral among the youngest generations for this reason.

So you should read the post if you’re building in consumer, but a few interesting things to pull out:

• Temu, the gamified ecommerce marketplace, was the most downloaded app of 2024. 

• There’s a new product catching fire in short form entertainment called ShortMax that’s essentially a slamming together of tik tok and netflix. Scripted hyper short form dramas in swipeable UX.

• Threads was the most downloaded Meta product and is significantly outperforming twitter.

• ChatGPT and an AI study companion called Gauth (developed by ByteDance) led in AI.

• All the traditional streaming apps are lagging, but need to look at usage to know if that’s just because they’re all already downloaded.

Dopamine, dopamine, dopamine. Yikes.