Dialogue with DefiOasis: Grasping multiple Alphas such as ai16z, how to develop AI Agent blue-chip hunters?

Interviewer: Wuyue, Xianrang

Interviewee: DefiOasis

This article is a written record of our interview with the Twitter AI Agent Blue Chip Hunter DefiOasis on January 26th. DefiOasis is the author/analyst of Wu Blockchain, focusing on on-chain data and investment research analysis on a daily basis. During the hot period of the AI Agent sector, he targeted multiple blue-chip targets, grasping ai16z, ACT, PIPPIN, TURBO and other targets with 10 times or even dozens of times returns. As of the interview, the overall return has reached 7-8 times.

In this interview, DefiOasis briefly popularized his investment experience, research methodology, and portfolio building in the AI Agent track, systematically sharing his valuable experience and insights, which are of great reference value for investors and practitioners.

1. Wuyue: Can you briefly review when you started investing in AI Agent-related Tokens?

Defioasis: I am usually interested in on-chain data analysis and on-chain asset-related gameplay. Since last year, my main focus has actually been on the on-chain sector. The motivation to research AI Agent came from my long-term exploration of on-chain assets. I have been observing on-chain assets since last year, but I have invested less real money, mainly because I feel that on-chain assets, especially memecoins, cannot convince me to invest a large amount of capital.

But the turning point was in late October to early November last year, when the phenomenal project Goat created the narrative of AI token issuance, and $GOAT soared to a market capitalization of several hundred million dollars in a very short period of time, and then Binance launched its spot and futures. This event made me rethink the AI-related track. Of course, there were some other events that had an impact on me before that. One was Worldcoin, whose issuer is Sam Altman, the co-founder of OpenAI. Worldcoin reached a fully diluted valuation of over 100 billion dollars under the narrative of OpenAI's Memecoin.

The second was $TURBO, a memecoin created by AI, which surged 200-300 times on the CEX. From these two events, I felt the strong interest of the public in the AI+Crypto theme, which also convinced me that the targets related to AI Agent are worth laying in wait.

So after $GOAT was listed on Binance, I quickly saw the potential of the AI Agent track and felt that Binance had the ability to lead this track, so I started to spend time searching for some AI-related targets. At that time, there were a few targets that caught my eye, one was ai16z, and the other was ACT.

In retrospect, ai16z seems much more powerful than ACT, but at the time, Shawn, the founder of ai16z, was not very famous, and there were also issues with the way he issued Tokens and the Token contract. Due to the large fluctuations and relatively small popularity of $ai16z, although I also built a position in ai16z at the time, the position was not large, and I later did not dare to buy more after $ai16z once fell below 10 million dollars in market capitalization. Instead, I heavily positioned in the more stable and larger-scale ACT.

The first time I bought $ACT was around early October last year, which was my first formal position-taking operation in the AI+Crypto track, buying about $3,000 worth. Later, when ACT reached a market capitalization of 20 million dollars, I bought it multiple times, with an average price of about $0.022, for a total of more than $10,000. What happened later was unexpected, ACT was actually listed on Binance, and I remember I was attending a conference in Bangkok when I learned this news, and I was quite shocked. After $ACT was listed on Binance, its fully diluted valuation reached 7-8 billion dollars, and I sold off a large portion of the Tokens the next day after Binance listing, and still hold some of them.

Subsequently, I did observe that the AI Agent can be said to be the only track that has developed from a generic Pump.fun to a vertically formed scale, so I started to delve deeper into this track for some investment research. After ACT, I also managed to grasp some other good targets, such as PIPPIN, which I once heavily positioned, achieving a return of over 10 times. In the end, my overall investment return in the AI Agent track from October to the last few days is about 7-8 times, with some pullback recently due to the market downturn, but still around 5 times.

2. Wuyue: So what is your research methodology for AI Agent? Can you give some specific cases to help everyone understand your research methodology?

defioasis: Many of the current AI-related assets are very different from the previous AI projects. Most of the assets that people are focusing on now are based on the fair launch of Pump.fun, and the assets held by the project parties or founders may not be as much as some big holders. In this case, it is necessary to pay attention to whether the project party has a long-term mindset, and whether their background conditions can support the project. If the founders have no morals, they may just abandon the project or start a new one.

So my methodology for AI Agent is to start from the people and examine the project, see if the founders really want to build the project and whether they have the ability to do it well. I can give an example using PIPPIN. In December, Solana and ai16z co-hosted an AI hackathon, and I noticed a Japanese person named Nakajima Yohei on the panel of judges and guests. I learned that he was doing an AI Agent project, which is PIPPIN, a child-oriented AI Agent project, which I found quite interesting as I hadn't seen a similar project positioning before.

Further research on Nakajima Yohei revealed that he is also the founder of Baby AGI, a project that has over 20,000 stars on Github. I then searched for more information and found that this BABY AGI is quite powerful, as a product of the AGI concept, it has been mentioned in many media and papers. I feel that this can prove the strength of this person.

In addition, this person is also a partner of a VC, and they have invested in some Web3 projects, some of which have been top-tier. So from the overall perspective, PIPPIN's founder is very powerful in terms of technology, capital operation, and backed resources. As a relatively prestigious person, the probability of him rugging or abandoning the project is much lower.

At the time, I saw that PIPPIN's market capitalization was around 20 million dollars, and not many people around me were paying attention to it, which was a target that had not yet been discovered in value. I tend to buy Tokens in the 10-20 million dollar FDV range. Later, I spent a total of over $40,000 to buy about 0.2% of the total PIPPIN supply.

Later, PIPPIN's FDV once fell below 10 million dollars, but I did not abandon my position because I felt that the founder's background and technology were good, and the fundamentals were still sound, and the short-term price fluctuations would not affect the long-term value of the project. Later, PIPPIN underwent a transformation, shifting from a single Agent to an AI framework, which led to a sharp increase in its valuation. After this news came out, even without a mature and deployed framework, the market recognized the founder and the project itself, and PIPPIN's FDV quickly soared to 200-300 million dollars.

Developing an AI development framework essentially means a split, and the market pricing for this type of Framework is the highest in the AI track, so I felt the project was good and held on without moving, and when the FDV reached 300 million dollars, I sold off a large portion, and the remaining part has not been touched until now. I feel that when its framework is truly formed and a split model is formed, it may reach a FDV of 1 billion dollars or even higher.

3. Wuyue: Which AI Agent-related projects are you currently optimistic about and why?

defioasis: In fact, there are still quite a few projects that I am bullish on, like PIPPIN, which I still hold. I generally choose targets with an FDV around $20 million, and there are other aspects as well. I used to often go prospecting in the AI hackathons of the Solana ecosystem, where there are quite a few award-winning projects, and I have been gradually screening and researching their fundamentals. Another one is Virtual, which is about to enter the Solana ecosystem, and there may be some interesting projects developed later. Virtual has already proven its success on the Base chain, so it is worth paying attention to.

Next, let me talk about some of the projects that came out of the Solana hackathons, but I won't provide any investment advice. Recently, I've noticed a project called Agentipy, which is also doing an open-source AI Agent framework, with the core being to connect AI Agents to Solana's chain applications using Python. Its roadmap shows that it may launch a self-narrating trading bot in Q1 and a LaunchPad in Q2. Most importantly, Agentipy mentioned that they will use their token APY as a flywheel to participate in it, so I also studied their token APY and think the design is not bad, although it is also based on a fair launch.

But the team behind it took 40% for themselves and all of it is subject to a two-year linear unlock. I think the fact that they can do this shows that the team has a certain determination, including the fact that Agentipy's co-founder and CEO have been followed by the . Plus, this project came out of the Solana hackathon, so it at least has a backing, and is relatively reliable. Of course, this project is still in a relatively early stage and has a lot of uncertainty. I will also pay more attention to the projects that are cross-chained to . There are quite a few opportunities in this area.

If we look at it as a whole, the AI+Crypto track is gradually transitioning to the AI Application stage, and after the various development frameworks have exploded, we need to pay more attention to AI+Application projects. I think AI+ is also not bad, that is, the native narratives of AI+Crypto as well as assets and Frameworks, there may be some good opportunities, but this area is still in a very early stage, and I haven't seen any good targets. I'm still keeping an eye out, and I haven't acquired any new assets recently, that's about it.

4. Wuyue: How do you view the current AI Agent track and its market status? How long do you think the AI Agent hype can last? Do you think the bubble has peaked?

defioasis: It has indeed been quite cool recently, but I think the opportunities for AI Agents are not over yet, because the AI-related concept is a relatively make-sense thing, and AI products outside the circle are still iterating rapidly, and are on an upward trend in both technology and capital, which I think is a more important foundation. In fact, many of the AI targets have been driven by factors outside the circle, such as Shawn, the founder of ai16z, who was probably a marginal figure in Web2, but seized the opportunity in to establish ai16z, becoming a leader in the AI+ track. I believe that under Shawn's influence, more talents from Web2 and traditional industries will enter to do things.

Furthermore, I think circle's AI projects are lagging behind the traditional industry, and many rumblings in the outside world, such as the big moves of AI giants, will be transmitted to the internal, forming new narratives and new sub-tracks. In addition, AI Agent is currently the only one that has transformed from a generic Pump.fun to a scalable vertical track.

Apart from AI Agent, there are actually not many projects that can break out of the simple memecoins, and the other one that can barely break out may be Desci, but it's also very cool now. Apart from that, there are no other tracks that can transform from generic to vertical, which in itself shows that there is actual demand behind the hype of the AI Agent narrative. The current overall AI Agent market has seen a significant pullback, mainly because it was a bit overheated before, and now the individual Agents are very rampant, which has forced everyone to get involved in the development framework, gradually becoming fatigued. Later on, it may be the era of AI plus applications, AI plus some native narratives. If my point of view can hold, there will still be some new hot spots and new opportunities.

5. Wuyue: The last question is actually about investment and trading experience, such as your portfolio building and exit strategies. Are you willing to share with us?

defioasis: Previously, we were talking about investment targets, but I think position management may be more important, because the selection of targets is essentially about looking at the project's technology, resources and background. Although it is now based on Pump.fun for fair launch, the research approach I think is not much different from VC coins, it's just about looking at the technical resources, team endorsement, chip structure analysis, rat race and blue-chip addresses, etc. I will focus on the issue of position management later.

Assuming there is a good asset, it will generally go through three stages: PVP, second stage consolidation, and listing. Most projects fail in the PVP stage, maybe just a few days and they're gone. And my past portfolio building style has mainly targeted assets that have gone through the second stage consolidation. For example, I choose targets with an FDV between $10-30 million, which I usually call the "on-chain sweet spot".

I found that some good targets, after spiking a couple of days before launch, will mostly fall back to this position and then go through some oscillating consolidation. I usually only make some position allocations at this stage, especially for targets that have experienced a 70% or more pullback and can stabilize around the $10-20 million FDV range, I will pay extra attention to them.

If I see a decent asset, I will first add it to my Watchlist, and then rate these projects, roughly divided into S-level, A-level or other levels. S-level are those that can form a pattern cycle , like those that can constantly output sub-s for profit as a mother ecosystem, and the sub-s can constantly create a wealth effect, and people need to buy the mother for transactions, thereby constantly generating demand for the mother , which is actually very similar to the Pump.fun and the mother flywheel of Sol and .

So most of the S-level ones are framework-type targets, because they have the opportunity to form an ecosystem or this kind of pattern flywheel. Why can PIPPIN rise? The main reason is that the market's expectation for it to do the framework is relatively high. As for the A-level projects, the main judgment is based on the narrative background, technical resources, team and other indicators, such as the founding team's past background is not bad, has the background of the Solana Foundation, or the founder has done some products that have attracted a lot of attention, and may develop into a new track, etc., I may give it an A-level, anyway, they are all added to the Watchlist first for observation.

Generally speaking, as long as they are S-level or A-level targets, I will build positions, each time around $2,000-$3,000, but whether to add positions or concentrate positions will be more cautious, requiring more time to observe, to decide whether to add positions, I need to go to the project's community to see what the developers are doing every day.

After setting such a standard, I also have a certain limit on the holding amount of each , such as I will only hold up to 0.2% at most, if the FDV is around $20 million, the purchase of 0.2% will cost about $40,000. In fact, each time I add positions, I will re-think the target, if I feel that the target is inconsistent with the original purchase logic, I will decisively give up adding positions, or if the target falls sharply below the $10-20 million range, I will also decisively give up adding positions.

In principle, I will sell out at least 10x, around the $200-300 million FDV range. Fundamentally, portfolio building is to ensure that I'm on the car, cautious about adding positions, gradually increasing the position, including why to set a 0.2% hard cap, it's all to avoid being trapped in a target due to excessive blind confidence by continuously adding positions.

Of course, as long as I've added positions, I'll believe in my judgment of the target. But the above is the portfolio building logic of the second stage consolidation, I feel that the market is also quite familiar with the routine of the second stage, but if you want to hold, it's still quite difficult to play, so I'm actually also doing some lottery-style plays, that is, the so-called PVP, but at this time I'm no longer limited to AI, as long as it's an on-chain asset, I'll get involved.

In the absence of a major market trend, I think this may also be a good choice, in a sense it is also preparing for the potential future major market trend, and I feel it is very important to always maintain this kind of profitable mindset, this is roughly my situation.

Wuyue: It sounds like you've gained a lot today, thank you very much for coming to have this interview with us.