GM friends, in a bull market, we’re always bombarded by the successful 100x on our timeline [insert Hyperliquid shill pls stop], but we’re not talking enough about the opportunity being missed.

In this piece, I present a retrospective on the story of Virtuals Protocol, how I met with the founders early on, and how I missed my first 100x as a VC.

Note: Technically, my first 1000x miss was the Solana seed round through an angel friend in 2019, but I wasn’t an investoooor then.

Disclaimer: I’m an Investor at @primitivecrypto (PV). PV is not a VC but writes VC checks. The opinions on this piece are solely mine.

Virtuals is my biggest liquid venture miss of the cycle. The founders first reached out to me in July around ETHCC when their FDV was ~$50M. Before that, I’ve even personally heard of the project from mutual friends since Q1, when they were valued even lower. Fast forward six months later, the AI agent tokenization platform is a main character in the current crypto cycle.

Jansen and Wee Kee, the co-founders of Virtuals, truly have that dog in them.

I remember they were tirelessly sharing about Virtuals with investors and operators in the space. Being also based in Southeast Asia (SEA) most of the time, I’ve heard from fellow crypto friends about their rebrand from the PathDAO era and their thesis on tokenization of AI agents. This sheer tenacity to continue working on a project even after experiencing the bear market, and without any significant CEX listing, is an impressive showcase of grit. Many other founders would’ve returned the capital or abandoned their projects given the same situation. The Virtuals team pushed through, and they came back stronger than ever.

What Led to My NGMI Decision

Earlier this year, we saw a slew of crypto x AI projects that are trying to decentralize computing or inference. I probably don’t have to say that there was much vaporware from those cohorts. The majority of those projects didn’t have a real way for retail users to participate. Sure, you might be able to get a few airdrops here and there by joining the network and running a few computations, but there was nothing as tangible as the GameFi or DeFi pool2 that allows for a scalable level of retail mania.

Initially, I had thought that this was going to be served by some sort of gamification in collecting unique data, combined with consumer-centric applications that make the experience unique — perhaps with some sprinkle of ponzinomics too. After all, data is the still bedrock of any AI model; and what better way is there to incentivize people to share unique data than free magic internet money?

Remember Westworld season 3?

Turns out, the crypto market shows that it’s continuously becoming more barbell. We skipped whatever I wrote in the last paragraph and jumped straight into asset issuance — which remains the number one crypto PMF.

The Virtuals team, given all of their prior hard work, was extremely well suited to capitalize on this opportunity.

Enter GOAT

They say luck is what happens when preparation meets opportunity.

At this point, I don’t need to explain what GOAT is; if you’re not aware of the phenomenon, this is a good explanatory article. In short, GOAT triggered the AI agent coin mania as it opened up the market’s imagination as to what’s possible when an AI agent can interact with a form of money. There are a lot of details here around GOAT’s limitations, how there’s some level of human intervention, etc. — but the key point remains the same, GOAT convinced people that when an AI agent meets crypto, it opened up a whole new world of experimentation.

Realizing this opportunity, the Virtuals team knew that they had to showcase their own technology and be quick about it.

LUNA, their own tokenized AI agent was launched on October 16th, roughly one week after GOAT was launched. If you’ve been in crypto for quite some time, you know being a beta to a main character is not enough. Over a breakfast session in Bangkok, Jansen told me that the Virtuals team sprinted hard to make Luna the first AI agent that could autonomously make an onchain transaction.

Learnings So You’re GMI

By its nature, a reflection is often subjective, but here are some of my takeaways:

• Tenacity: The Virtuals team has showcased an exceptional level of tenacity by continuously iterating on various products. Different founders have different motivation structures based on their personal reasons and socioeconomic background. Regardless of those factors, it’s key to bet on teams that don’t give up, maintain an intense level of ambition, and stay on top of market trends.
• Geo-Arbitrage & Rapid Experimentation: Generally, projects that try to do rapid experimentation (more of a platform/launchpad-based model) have a hard time excelling in the Western markets due to the high-cost structure. Being based in SEA, Virtuals were able to rapidly iterate with a lower cost structure but get the benefit of a USD-denominated user base and capital market.
• Resilience & Business-First: The strength of SEA-based founders is resilience. The region has always been about business-first, taking what worked in the Western/Chinese markets and implementing it locally, whether you’re talking about conventional, web2, or crypto businesses. This culture of not being an ideological or technological maxi, but a business-first mindset, was strongly demonstrated by the Virtuals founders. They knew what they didn’t know, what they were lacking, and what’s their moat.

What Comes Next

The AI agent mania has only been happening for ~2 months but I’ve felt like I’ve aged for two years. While the market is already showing some signs of fatigue, I believe we’re due for more crypto x AI mania in 2025. Crypto vertical always starts at the most degenerate corner, before evolving into more mature use cases.

One thing remains true → Without crypto, AI experimentation is severely limited.

This is most obvious with AI agent experiments. You can’t let a random AI agent touch real capital in the TradFi system without stacks of paperwork and lawyers behind your backs. You can’t give a random AI agent literal cash. Crypto is the purest form of money for a digital sentient being.

Thus, AI agent experimentations will evolve from simply a GPT wrapper that can tweet trading at $100M FDV to more interesting use cases — personally, I’m excited to look at:

• More AI agent tokenization frameworks and tokenization platforms. While the Virtuals team is consistently shipping new products, there is still room for competition. Other platforms such as @ai16zdao, @MoemateAI, @Spectral_Labs, and @griffaindotcom have emerged to take some market share.
• Niche AI agent experimentation. Projects such as @freysa_ai, @aiwdaddyissues, and @Big_Pharmai shows that there are more niche experiments and use cases we can explore. The key is how you evolve from being just a fun experiment to a legitimate protocol.
• Consumer crypto x AI agent. How do you keep the benefit and novelty of AI agents, but package it in a way that’s enticing as a consumer app? It can even be merged with other AI products such as data, training, or inference — the key is to maintain the novelty of AI agents.
• AI agent x vice. I don’t dare to pontificate too much here. But we’ll see more AI agents who are making significant cash flow not simply because of its token or trading the crypto markets but by entering into vice businesses (gambling, adult industry, etc.)
• AI agent x payment. Advance agent-to-agent interaction and how we can use agents to unlock a more seamless payment experience offchain and onchain.

VC mindset can sometimes be flawed when it comes down to community-driven initiatives. The key to learning is to always be open to new experiments, unshackled by conventional norms, and adapt, not idealize. Primitive is always on the lookout for daring founders, if you’re working on any of the above, let’s chat!

Disclaimer:

1. This article is reprinted from [X]. All copyrights belong to the original author [@ManoppoMarco]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.