GuShiio.com This interview discussed in depth the investment trends, market status and future opportunities of AI Agent projects.
edit| Wu said blockchain
This interview discussed in depth the investment trends, market status and future opportunities of AI Agent projects. Analyst defioasis shared his experience in investing in on-chain assets, memecoin and AI-related projects, including the rise of key projects (such as GOAT, WorldCoin, Turbo, Pippin, etc.) and the market logic behind them. In addition, it also analyzes investment methodology, position management strategies, signs of cooling market bubbles, as well as the importance of changes in industry trends and project narratives.
Please briefly introduce yourself, and review when you started studying AI Agents, when you officially started investing in it, and what is the overall return rate at present
Defiosi: Hello everyone, I am defiosi. I am usually interested in the assets on the chain and the derived gameplay. Since last year, my main focus has been on playing on the chain.
In fact, the research on AI Agents also stems from mining on the chain. I have been observing assets on the chain since last year, but the pure Memecoin is actually not that good at it. I mainly watch more and I don’t particularly convince myself to play with big money.
The turning point was mainly in late October and early November last year, when GOAT soared to hundreds of millions of dollars in a short period of time and Binance’s launch of GOAT futures made me rethink the track. Of course, before I talk about some AI-related things before GOAT. Before GOAT, there were two major events. One was WorldCoin. Because Lianchuang was Sam, the founder of OpenAI, WLD was regarded as OpenAI’s Meme in Crypto. The FDV once exceeded US$100 billion; The second was Turbo, which claimed to be an AI Meme created by GPT. It was also the target that increased 200-300 times in CEX last year. So these two things show that everyone is still very interested in the theme of AI+Crypto and worthy of hype.
Back in late October last year, after GOAT joined Binance Futures, I quickly thought of WorldCoin and Turbo, and Binance had the ability to lead the track. At that time, there were four assets that I found very eye-catching. Two of them were AI-related, one was ai16z and the other was ACT; the other two non-AI-related assets were LUCE and BAN. Looking at it now, ai16z seems to be much more powerful than ACT, but the situation at that time was that the founder of ai16z, Shaw, had only nobody, and ai16z was launched using daosfun. There was a problem with the creation of token contracts, and the fluctuations were particularly large at that time because the pool was very small. At that time, I opened a position in ai16z at a market value of US$25 million, which cost about US$3000, but the pool was very small and the market value of US$10 million was once due to fud, so I didn’t dare to add a position. So I finally chose a more stable ACT with a larger number of holders. ACT can be regarded as my first operation to officially open a position in AI investment. The first purchase was on November 5th when I bought the first $3000 ACT. Later, I found out that there were more than $10,000 ACT on Bitget with a market value of $22 million and an average price of $0.022. But at that time, I didn’t expect that I hadn’t bought enough positions after buying one after another, and then I was surprised to buy the coins. I remember that I was still attending the conference in Bangkok, but I was actually quite shocked. After joining Binance, the market value reached a maximum of US$700 – 800 million, but I actually sold most of it at US$500 million the next morning after the Binance news, and the rest have not been sold yet.
It was also observed later that AI Agents was the only track that went from universal PumpFun to vertical and formed a scale, and later went deep into this track to make investments. After the ACT, some good targets were caught, such as Pippin, which was also heavily stocked and achieved a return of more than ten times. The overall position used to invest in AI alone has actually reached 7-8 times since November, and has recently retreated a little, about 5 times.
What is your approach to studying AI Agent projects? Can you give a case study of the deepest AI Agent project you have studied so that more people can understand your research methodology
Defioasis: Today’s AI agents or some AI-related assets are many different from previous AI projects. Nowadays, the basics are assets based on fair launch based on PumpFun launches, so the project party or founder may not have many assets in their hands, which may be even less than some snipers or big players. This means that the founder’s character, philosophy and background are very important. Otherwise, if it may not have morality or a real name, it can abandon the title at any time and open a new title. So my methodology for AI Agents starts from the perspective of people, whether they really want to do it, whether they have the ability to do it, and what they can do.
You can give an example, take the experience of Pippin’s project as an example. Yohei Nakajima was first known on December 11 from the judges of Solana AI Hakathoh, a hackathong event organized by ai16z and Solana. At that time, I looked at the guest list of the judges and found a Japanese Yohei Nakajima, the founder of Pippin. It was quite interesting to see him working on a child-oriented AI Agents, because I had never seen agents with the same type of positioning. Another aspect is that being a judge is obviously more valuable than being a judge who still needs to participate in competitions to prove themselves. Later, further research on Yohei Nakajima found that he was the founder of BabyAGI, and there were more than 20,000 stars on BabyAGI’s Github. At that time, I searched some information and found that BabyAGI was quite powerful. As an AGI concept product, it had been quoted in many media papers, which also proved its strength. In addition, Yohei Nakajima is himself a partner of Uncapped Capital, and his main Web3 projects are Pixel, which is on Binance. On the whole, Yohei Nakajima, the founder of Pippin, is very powerful in terms of both technical and capital resources. As a person with a real name and certain prestige, the probability of him rugging casually is much lower.
At that time, Pippin’s market value was about US$20 million, but not many people paid attention to it at that time. A market value of US$20 million is also in line with my own requirements for purchasing the underlying asset, because I prefer to buy in the market value range of 10-20- 30M. I successively bought two-thousandths of the total amount, which is what I define as the single maximum position limit, and spent about US$40,000. He actually fell below $10 million later, but I bought it with chips. For example, the founder’s background and technology will not change with price fluctuations. So I didn’t care about it after collecting it.
Later, Pippin announced that it would transform into a framework, from a single AI Agent to an AI Framework. This valuation is a straight increase. At that time, after this news, even though we have not yet worked out what the framework looks like, everyone’s recognition of the founder’s technology and capabilities also caused Pippin to soar to US$300 million. Making a framework means that it is possible to become a split. The market price for the framework is currently the highest on the ai agents track, and the framework or ecosystem that can form a split model has the opportunity to exceed the market value of more than US$1 billion.
Which AI Agents projects are you optimistic about? What are these projects for? Why watch?
Defioasis: There are many optimistic ones, like Pippin, but I don’t want to talk about it if the market value is relatively high. I still usually choose around $20 million, but there are actually fewer people who will place heavy positions.
At present, there are two main directions. One is the gold rush in the Solana AI hackathon. It is now over. There are quite a few award projects in it that are being screened recently. The other is that Virtuals went to Solana and collaborated with Jupiter. There will be many interesting projects later, because Virtuals has already proved its success on Base. Of course, I’m still looking at this one.
Here is mainly about some projects released by Solana hackers. Let me share one here, but this is not investment advice. A project I was observing recently, AgentiPy, is actually an open source framework that uses python to connect ai agents to solana’s on-chain applications. According to the roadmap, autonomous narrative trading robots may be launched in Q1 and launchpad in Q2; the most important thing is that its token APY will participate as a flywheel. I studied APY token economics and designed it well. Although it is also based on a fair launch like PumpFun, it itself played 40% of the chips and placed all of it in Streamflow for two-year linear unlocking, which shows that the team still has a certain determination. AgentiPy’s joint innovation and CTO have been promoted by Solana officials. After being released from Solana hacking, it is at least an endorsement. Of course, the overall situation is still relatively early and there is relatively great uncertainty. Later, I will pay more attention to the projects launched on Virtuals after cross-linking to Solana.
Looking at the overall situation, I think AI is slowly reaching the stage of AI Application. In addition to the framework, AI+ applications will also keep an eye on it later, especially AI+DeFi. This is the combination of AI and Crypto’s native narrative, as well as DeFi assets and flywheel. There may also be some good opportunities, but it is still in a very early stage, and I didn’t see any good targets. I still keep observing now, and I haven’t acquired any new assets recently.
What are your views on the current AI Agents track and market status? Do you think the popularity of AI Agents will last for a long time or do you think that this bubble has peaked?
Defioasis: It has been cooler recently, but I don’t think it will end yet. AI is still a matter of making sense. AI outside the circle is still rapidly iterating and developing, and it is on the rise of technology and capital. This is a more important foundation. In fact, many AI targets are actually driven by the outside world, whether it is narratives or talents. Wasn’t Shaw also a fringe person on web2 back then? The ai16z he created now has become the leader in Crypto AI. Under Shaw’s influence, I think there will still be more technical talents from traditional industries coming in to do AI things. Crypto AI itself is significantly lagging behind the circle, so any disturbance outside the circle and updates at the Dabie level will be transmitted to Crypto to form a new narrative and new sub-track.
From another level, from the perspective of the circle, AI Agents is the only one that has formed a large-scale vertical track from universal PumpFun. DeSci may be half of it and has been relatively cool recently. In addition, there is no other track that can move from universal to vertical, which actually shows that there is a strong demand for AI narratives. AI Agents are relatively cool now, which is to cool down the previous overheating. Nowadays, the proliferation of single Agents forces everyone to make frameworks and roll, and everyone is gradually becoming more tired of this kind of thing. Therefore, if AI+ applications, especially the AI+Crypto native narrative, can be established, I believe it will re-drive a wave of craze and new opportunities.
Experience about investment and trading, and what are the tips for opening a position and exiting
Defioasis: The above actually talk about the issue of investment targets, but actually I think position management is more important. The selection of targets is nothing more than technology, resources and background. Although it is based on PumpFun for fair launch, the research ideas are not much different from VC coins, such as technology, resources, background, team, endorsements, etc., as well as analysis of the chip structure, rat warehouse blue chip address, etc.
Later, we will focus on the issue of position management. Nowadays, a good asset can basically go through three stages: PvP, second paragraph and Shangsuo, but most PvP is gone. I usually mainly play the second stage, focusing on laying out assets in the market value range of US$1,000 -2,000- 30 million, which I also call the sweet spot on the chain. Because I found that some good assets will probably consolidate and fluctuate at this position after they are born and rise. I usually only allocate assets within this range. Those who have experienced one or more retracements of around 70% and are relatively stable within the market value of US$1,000 -2,000- 30 million will pay attention.
When I see a good asset, I will add it to the watchlist, which will be divided into S-level, A-level and others. S-level is a coin that can form a model. For example, it serves as a parent currency ecosystem and can benefit from the continuous production of sub-coins. For example, as a transaction on assets, and the sub-coins create wealth effects, this parent currency will be needed to conduct transactions and generate constant demand. This is the same as PumpFun’s SOL and Virtuals ‘parent currency flywheel. Therefore, most of the S-class are frame types and have the opportunity to form ecological or model flywheels. For example, why Pippin can get up is because the market has high expectations for it to frame.
Level A is mainly based on narrative, background, technology, resources and team (which cannot form a model). The past background of the founder and founding team is very important, such as the background of the Solana Foundation, the existence of many Github stars, the creation of products that have attracted attention, the narrative is not bad and the possibility of developing into a segmented track, etc.
Generally speaking, both A-level and S-level positions will be opened, and each time a position is opened is about 2- 3,000 U. But I will be more strict about choosing whether to add a position or heavy a position. I need to observe for a period of time and decide whether to add a position. It will mainly depend on what the community and developers are doing on a daily basis.
I also have certain standards for the amount of single currency held in each currency. The maximum holding of a single coin is 0.2%, which means that the total cost of purchasing is approximately 40,000 U if it is around the market value of US$20 million. Every time you choose to add a position, you will think about the target. If it is judged that the target is contrary to the original purchase logic, you will decisively give up adding a position, and the remaining positions will be temporarily retained; if you fall significantly below the box market value range, you will decisively give up adding a position, and the remaining positions will be temporarily retained. In principle, if you get at least 10 times before considering whether to sell, it will be considered in the market value range of 200-300 meters.
Often, the target for opening positions is to ensure that they are on the vehicle. The purpose of carefully adding positions and gradually placing positions in batches, and setting a hard cap is to avoid adding positions all the way due to excessive blind confidence and allowing funds to be trapped in a certain target. Of course, as long as you choose to add a position, you will believe your judgment on this target.
All the above are some two-stage gameplay. I feel that in the market, whether users, Dev or coin issuing factories seem to be more familiar with the routines of these two paragraphs, and I feel that the recent two paragraphs have been relatively difficult to play. So now I am also doing some lottery methods to play a wave, which is PvP. At this time, it is not limited to AI. I have prepared several SOLs, each time 0.1 SOL to buy, watch, monitor and follow some orders to select addresses to make a small profit. I have felt in the past two days that in this market, it is easier to make profits by lottery streaming games. This may be a probability. With so many plates, as long as the investment is small and scattered enough, and some high-winning addresses are monitored as an assistance, the lottery flow can recover all previous losses with one win. In this lack of big market conditions, it will be a good choice, and to some extent, it will also prepare for a potential big market. It is important to always maintain the feeling and mentality of profit.
