Alliance Dao discusses the trend of AI and cryptocurrency integration: Where are AI agents headed?

Compilation: Wu Says Blockchain

In this episode of the "Good Game Podcast," Imran and Qiao discuss the intersection of artificial intelligence and cryptocurrency, including market cycles, the use of AI agents to replace KOLs with governance tokens, and the impact of stablecoins in developing countries. They also mention the role of DePIN in coordinating industries such as energy and computing, and highlight the potential for seamless integration of AI and cryptocurrency. While optimistic about innovation, they also caution against excessive hype.

Challenges of DeepSeek's Self-Funding Model to Traditional AI Startups

Imran: Last time I was in Dubai, the whole area was flooded. I had to switch from one hotel to another. But during that time, I had a lot of thoughts because I feel we are entering a new era of cryptocurrency. Many things seem about to change, right? Like today’s announcement from Mark Zuckerberg. Did you see it?

Qiao: You mean he fired the entire content moderation team?

Imran: Yes, and moved some of them to Texas. There is a clear bias in California. So, I think the next four years will at least be a completely different era from the past few years. This will bring new business models and technological updates, such as developments in AI or innovations in cryptocurrency like stablecoins. We are really entering a new phase. It’s interesting. Did you see that update… something called DeepMind or DeepDive? A team of Chinese engineers successfully defeated OpenAI.

Qiao: That was last week.

Imran: Right, it was last week. You know what I'm talking about, right?

Qiao: Yeah, I know.

Imran: I think this is a big event. Now, you don't need to spend hundreds of millions on hardware. You can launch competitive products at just a tenth of the cost. Do you know the details behind it? I haven't looked into it deeply myself.

Qiao: No, I just know they surpassed OpenAI in various benchmarks. But I'm not sure if it has to do with costs.

Imran: I think part of the reason is they didn't raise any venture capital. It’s said that this team is completely self-funded. I read that an AI expert from Silicon Valley mentioned that future AI startups might not need venture capital at all. These people come from Citadel or some quantitative trading team, and after leaving, they start their own businesses, completely self-funded. This is interesting because similar things are happening in the cryptocurrency space, right?

Qiao: Wait, they did it without venture capital?

Imran: Yes. They bought a bunch of hardware. I remember they used their own funds to purchase $6 million worth of hardware and then built this project. What was it called again? DeepSeek… hold on, I have a bookmark, let me check…

Qiao: I just read a TechCrunch article that said this is a well-funded AI lab.

Imran: I saw somewhere that they mentioned they didn't accept any venture capital.

Qiao: I feel like if you don't take venture capital now, you can't compete with OpenAI at all.

Imran: Maybe that was wrong information.

Qiao: You need a lot of data, hardware, and great engineers. Maybe the founders themselves are wealthy, or they have special liquidity?

Imran: DeepSeek is funded by a quantitative trading company, but not in the traditional sense of venture capital. So they do have funding sources, but it all comes from the quantitative trading company they left.

Qiao: So, they do have money.

Imran: Yes, they do have money. They operate using parametric models instead of relying on external venture capital.

Competition in the Commoditization of Base Models and Distribution Channels

Qiao: From this news, I think the moat for base models might not be that deep anymore. Companies like OpenAI, Google, Anthropic, and even Facebook's Llama and some Chinese companies, I feel they will converge at some point. At the base model level, this technology might ultimately be commoditized. Because as long as you can access all the data on the internet and have enough funding, this technology itself is no longer a secret weapon. Now it’s more about engineering implementation.

Imran: Yes, I agree.

Qiao: So I actually expect these base models to converge eventually. Actually, you know Fred Wilson, right? He writes a New Year prediction article every year.

His first point in the 2025 predictions is that Google and another big tech company—I forgot which one—will defeat OpenAI through distribution capabilities. What he means is that the technology has been commoditized, and distribution channels are key.

Imran: Exactly, that’s my thought. I think the most important thing now is to master the end user. If you have distribution capabilities, can directly reach users, and get them to use your product, that’s what matters. In fact, I recently watched a YC (Y Combinator) video about AI, and they mentioned that initially they didn’t think routing technology (AI Routing or Model Routing) would become an important field. Even so, they funded some startups in this area, and it turned out to be a huge growth direction.

Ultimately, if you can control the end users, you can switch between different models based on user needs. I remember this technology is called "Open Routing Models" or something similar. Essentially, it’s a router that can connect to specific models based on application needs. This also indicates that the models themselves are gradually being commoditized.

Qiao: Yes, I agree.

The Rise of AI Agent Hype: From White Papers to GitHub Transition

Qiao: So, what do you think about the current hype around AI agents?

Imran: I find it interesting. Just yesterday, I was discussing this topic with one of the founders of Slop in our team. I described it like this:

2017 was the era of ICO white papers. Anyone with a white paper, or even just a connection to Vitalik Buterin, could easily secure funding, and token prices would soar as a result. Then came the DeFi summer of 2020, where projects like SushiSwap received funding and quickly rose. After that, it was the era of NFTs.

And the era we are in now is very similar to those. The difference is that we no longer rely on white papers, but rather on GitHub repositories. People evaluate projects based on activity on GitHub and the number of Twitter followers. These have almost replaced the promotional methods of ICOs.

The market is currently filled with hype and noise, with many teams securing funding based solely on superficial signals. Of course, there are some truly strong teams building impactful products. However, due to the frenzy of the bull market, the high noise level makes it harder for trustworthy teams to gain real recognition in the market.

Signals and Noise in the Current AI Metaverse Market

Qiao: Are we nearing the end of the AI meta era?

Imran: I don’t think we’ve reached the endpoint of the AI meta era. My main basis is that just yesterday or the day before, Sam Altman tweeted about AGI. He mentioned AGI (Artificial General Intelligence) and the prospect of agents gradually entering the labor market. While he didn’t explicitly state that AGI would emerge in 2025, he did mention that robots would start entering the labor market.

This reignited optimism around crypto AI agents. Additionally, last night Jensen Huang gave an important speech introducing their upcoming new hardware. He emphasized the next major opportunity for AI agents and described it as a trillion-dollar market.

From these macro dynamics, I think people’s enthusiasm for crypto AI agents is reviving. However, so far, we haven’t found any truly successful application scenarios for crypto AI agents. Of course, there are some examples, but none have clearly stood out as successful products in the intersection of cryptocurrency and AI agents.

For instance, Sphere One is a platform where you can talk to an agent, and it helps you complete orders—this is interesting. Meanwhile, the Orbit team and the Slop team are trying some more experimental things. They are exploring how to get people to play games together and co-create videos. But I think their ultimate direction is to build a social media platform that works for both agents and non-agents.

These two teams are ones we are closely watching. Of course, there are many other interesting attempts in this field, and I could list them one by one. But for now, these are the highlights I see.

Potential AI Bubble: Parallels to the 1999 Boom and Bust

Qiao: From a macro perspective, I wouldn’t be surprised if we experience a bubble similar to 1999. In the next 2-3 years, I think there’s a 50% chance of this happening. This means that AI-related stocks will experience dramatic, exponential, and parabolic growth, eventually spilling over into the cryptocurrency space.

Imran: Do you think similar dynamics will occur?

Qiao: Yes, stocks might trade at ridiculous valuations, like absurd P/E and P/S ratios, along with generally inflated metrics. Additionally, it’s likely that AI infrastructure companies will become the most valuable players, just like we see today with Nvidia and TSM. This is similar to the tech bubble of 1999 when companies like Cisco and Intel were the "picks and shovels" providers for the internet, rather than application layer companies.

The application layer giants like Google, Facebook, and Amazon didn’t rise until the next decade. Of course, Amazon had already achieved $1 billion in annual revenue in 1999, but the biggest players at that time were still infrastructure companies. I wouldn’t be surprised if we see a similar trajectory in the AI space.

Imran: Why do you think it will take two to three years? Why not this year?

Qiao: It could happen this year, but I don’t yet feel that extreme frenzy in the AI space. There is some hype, but it hasn’t reached a "crazy" level. For example, Nvidia’s current P/E ratio is around 30 times. That’s very reasonable for a tech company growing at 50% annually.

Whereas in 1999, companies like Cisco and Intel had P/E ratios in the hundreds. Many large internet companies were not profitable at all, and some were even losing money. But today’s Nvidia and TSM are not like that. So, a bubble could emerge this year, but so far, the market hasn’t shown that kind of frenzy.

I think people often underestimate the time it takes for AI to truly go mainstream.

Imran: I agree with most of what you said. I do believe we will see a situation in the AI space similar to the internet bubble. However, I have some reservations about the exponential growth of AI models themselves. This growth drives better product development and meets user needs instantly. This could lead to dynamics different from the internet bubble. Today’s AI products have a direct impact on users, whereas many internet services were still in their infancy during the internet bubble.

Qiao: That’s true. But we also can’t overlook the immense value created during the internet bubble. For example, Amazon had already generated $1 billion in revenue in 1999.

Imran: Yes, but the internet’s reach was far less extensive back then. There were no distribution channels or delivery systems like we have today. And AI is already widely usable across various devices.

Qiao: Of course, there are many practical AI products now, like ChatGPT, which is very useful, and code assistance tools like Cursor are also valuable. But if you compare the current state to the potential market size people imagine for AI, it still seems relatively small. AI is expected to have a huge impact on society, but achieving that may take ten or even twenty years. We might see significant productivity gains during that time.

Imran: Yes, that makes a lot of sense.

Qiao: It will take at least ten or twenty years; it’s impossible to achieve this year. For instance, global GDP cannot grow by 50% this year. It might not even reach 5%, likely staying in the 2%-3% range. It may take ten years to achieve stable growth of 5%-7%.

This morning, I saw many people on Twitter expressing excessive concern about AI. Everyone seems to be panicking.

Imran: Indeed, the sentiment around AI has become somewhat overly enthusiastic.

Impact and Popularization of AI Technology: Evolution from Personal Life to Market Cycles

Qiao: I’ve been thinking about this for a while. Since ChatGPT was first launched two years ago, I’ve been trying to use AI and considering how it might change my life. My conclusion is: it won’t significantly change my life. I will continue to live as I did before. Maybe in ten years, productivity will improve, but my daily life won’t undergo huge changes. For example, my children’s education won’t change much either. I still want them to focus on independent thinking and creativity, rather than mere memorization and repetition. AI won’t change that.

Imran: I agree, huge changes won’t happen overnight, but AI has already brought some small changes to my life. For instance, I use AI to help my daughter learn some things. And for medical diagnosis, for example, when I was in Turkey, my son got sick, and I used ChatGPT to analyze symptoms and get possible diagnoses.

But widespread adoption depends on user behavior. As deep users of AI, we can quickly see its benefits. But what about others? That’s still an unknown. I recently got my mother to start using ChatGPT, and now she uses it frequently. However, when I ask others if they’ve used ChatGPT, most say no. Ultimately, it comes down to whether people’s behavior will change to adapt to these new products.

Qiao: ChatGPT already has over 100 million users.

Imran: Monthly active users are around 300 million.

Qiao: That’s about 5% of the global population. In terms of adoption rate, it has already surpassed cryptocurrency. However, I’m not that worried about AI. Yes, the world will see some changes, and it might happen a bit faster, but it won’t be a drastic, sudden shift.

Imran: Take the Optimus robot, for example. In their recent demonstration, most operations still required a lot of human assistance and weren’t fully autonomous. It’s a bit like Tesla’s self-driving cars. When I bought my Tesla in 2017, it only had partial self-driving capabilities. In the following four or five years, with improvements in data collection, it has now achieved full autonomy. But that process took about eight years.

I think it may take even longer for AI agents to operate effectively in the workplace, at least six to ten years. Sam Altman might feel competitive pressure from projects like DeepSeek, but to see real progress, it will take time.

Qiao: People often forget that AI is not a new thing. It has existed for decades. The advancements in data, algorithms, and hardware over the past 40 to 50 years have brought us to today’s state.

Imran: Yes, Eliza was one of the earliest AI chatbots, developed by Joseph Weizenbaum in 1966. It simulated conversations with a psychotherapist. So these concepts have existed for a long time; it’s just that the technology has finally caught up.

Qiao: Returning to the hype around AI agents, it all depends on whether AI stocks will experience a bubble similar to 1999 and how this interacts with the cryptocurrency cycle. As it stands, I think the AI hype can last at least a few more months. Just last week, people were debating whether AI agents had entered the "mid-curve" stage. I was worried we might be nearing the frenzy stage, but now I’m not so sure.

Imran: I think as long as AI companies like OpenAI and Nvidia keep releasing new news, this hype around AI agents will continue to drive developments in the crypto space. It could last far longer than just a few months. Just last week, I thought this hype was about to fade, but the significant announcements from Sam Altman and Jensen Huang reignited the enthusiasm.

End of the Bull Market and Future Possibilities

Imran: So, what stage are we currently in the market cycle? This is a question we often discuss to help the audience better understand market dynamics.

Qiao: I feel the current market cycle is roughly at the 8/9 position. In terms of price and time, there might be 10%-15% growth left, and perhaps 10%-20% time remaining.

Imran: That probably means there are 3 to 12 months left. There could be two scenarios: one is that prices rapidly enter a parabolic rise, pushing the market to a quick peak, with the cycle ending between March and June; the other is that the market experiences a period of consolidation before a parabolic rise occurs.

Qiao: That makes sense and aligns with market consensus; I have no objections to that.

Imran: Unless there are a series of new major announcements in the AI space, this trend may continue as expected. If AI continues to power the crypto market, it could drive the market higher.

Market Risks: Macro Factors, MicroStrategy Hazards, and Cycle End

Qiao: Key risk factors this year include Trump—will he deliver on the policies he hinted at, like establishing a Bitcoin reserve?

Imran: If he really does that, and other countries start to follow suit, the entire market cycle could be disrupted.

Qiao: Macro factors are also a major risk. Will inflation resurge this year? Will quantitative easing (QE) restart? How will interest rate cuts unfold? Currently, U.S. stock valuations are very high, close to the levels seen during the 2021 bubble, slightly below the peak of the 1999 internet bubble.

Imran: But the economic situation now is completely different from back then.

Qiao: That’s true, but the market is still expensive, which increases risk. Another major risk is MicroStrategy and Michael Saylor.

Imran: GCR recently discussed this topic on Telegram. He mentioned that currently, MicroStrategy has no issues, but if their average Bitcoin cost rises to $150,000, it could become a hazard. Right now, their cost is around $60,000 to $70,000.

Qiao: Why would their average cost rising to $150,000 be a problem?

Imran: It has to do with volatility. If Bitcoin prices can’t maintain that level, MicroStrategy could face issues. If the SEC or other external factors cause MicroStrategy’s stock price to plummet, it could trigger panic selling.

Qiao: Dan from CMS has a different view. He believes that because MicroStrategy’s stock price is at a premium relative to Bitcoin, people are arbitraging by selling MicroStrategy stock and buying Bitcoin. This method is effective, but if MicroStrategy’s stock price becomes a discount relative to Bitcoin, that situation could reverse and pose serious risks.

Imran: Exactly, it depends on the broader market context. A day or two of volatility isn’t a big deal, but if MicroStrategy gets into long-term trouble, it could trigger a chain reaction.

Qiao: Regardless, I feel the market is still in an upward trend.

Imran: Yes, but from the timeline on social media, it seems everyone has a bit of PTSD (post-traumatic stress response). Last week, some predicted Bitcoin would peak at $70,000 or $85,000.

Qiao: Indeed, we are now closer to the end of the cycle than to the middle.

Imran: If you’ve already made a profit, now is a good time to start cashing out gradually. No one can perfectly predict the top, so it’s wiser to lock in profits while you have the chance.

Qiao: I haven’t started cashing out yet.

Imran: Neither have I. But for our audience, this is our third or fourth market cycle, and we are used to walking the edge of the cycle.

Controversy Over Tokenization of AI Agents

Qiao: This morning I realized I kind of miss the bear market.

Imran: Same here. Everything was quiet back then; you could focus on building products and engaging with users. Now, everyone feels like a genius.

Qiao: I also find the AI hype not that exciting.

Imran: I was excited at first, but after seeing so many projects, I started to feel bored.

Qiao: My confusion lies in the fact that good products don’t need tokens, and good tokens don’t need products. If an AI agent is an excellent product, why does it need a token?

Imran: I mostly agree, but in certain edge cases, tokens might make sense. For example, an agent governed by tokens could be used to manage revenue or work output, leaving room for innovative governance and token design.

Qiao: That’s true, but I remain skeptical about whether this trend of tokenization will succeed in the long term.

Imran: That’s why we’ve only invested in a few AI crypto startups. We want to test the waters while remaining cautious. If the projects succeed, that’s great; if they fail, we can still learn from the experience.

PumpFun Launch: Memecoins, DeSci, AI Agents

Qiao: Is there anything else interesting happening?

Imran: PumpFun has launched a startup project launchpad, and recently a project called Hyperfy went live—this is a gaming metaverse ecosystem built around the Eliza framework. It’s open-source and integrates Eliza into the web.

Qiao: They’ve turned the fund into a launchpad covering various projects. People usually see Pump as a meme coin launchpad, but projects like decentralized science (DeSci) have also launched here.

Imran: Exactly.

Qiao: Most AI agent coins are released on Pump.

Imran: The Pump fund is becoming a "liquidity black hole" across all domains. It almost dominates capital allocation.

Qiao: Where have I heard the term "liquidity black hole" before?

Imran: Oh, that was coined by Ben; remember OlympusDAO (OHM)? That’s where the term originated.

Qiao: That analogy is indeed very apt.

AI Agent Hype and DeFi Summer

Qiao: I saw someone on Twitter comparing the hype around AI agents to DeFi Summer.

Imran: Have you seen the chart from Messari?

Qiao: Yes, I have. The core idea of the chart is that the market cap of the agent hype now resembles the early stages of DeFi Summer. In the year following DeFi Summer, DeFi activity exploded due to the overall market growth. So the chart suggests that the agent hype is still in its early stages. But frankly, my interest in the agent hype is far less than during DeFi Summer. The atmosphere back then was unparalleled.

Imran: DeFi Summer was indeed a magical moment. I think crypto Twitter wants to recreate that magic, which is why they are pushing these narratives.

Qiao: Exactly. DeFi Summer felt more like a "right curve" hype, while the agent hype feels more like a "left curve."

Imran: It really is a "left curve." I was expecting more "right curve" innovations, but some tweets have genuinely disappointed me.

Qiao: Is there anything else worth paying attention to?

Imran: We’ve already discussed how the Pump fund is becoming a liquidity black hole. Additionally, we’re seeing startups launching at an astonishing speed—white papers are going live on GitHub at lightning speed. This reminds me of 2017 when ICOs were a craze, and platforms like Token Relations and Ian Balina’s site were the go-to tools for finding projects. Back then, you needed to read white papers, research teams, and participate with ETH or BTC, which could take hours or even days. Now, you just need to see a code labeled "AI" and then go all in.

Qiao: And then sell it two seconds later.

Imran: You have to do that, or you’ll get dumped. That’s how the system works now.

Qiao: This is very similar to the evolution of social media, shifting from requiring long attention spans to short content consumption.

Imran: Absolutely right. This shift also affects capital allocation. Pump has become the de facto platform, just as automated market makers (AMMs) are the foundation for token swaps. Pump is now crucial for crypto AI startups.

Decentralized Computing Projects: io net, Helium, Glow

Qiao: Which crypto AI projects are you interested in?

Imran: I’m interested in decentralized computing projects like io net. They reportedly generate significant revenue from decentralized GPUs. For example, Helium has an annual revenue of about $10 million, and io net is performing quite well too.

Qiao: I’m skeptical about the decentralized GPU market. I think in the long run, decentralization won’t provide enough economies of scale to compete with centralized providers. Centralization will ultimately offer cheaper services.

Imran: That’s a reasonable point, but I’m curious about who is using these decentralized GPUs today. Regardless, the leading DePIN (Decentralized Physical Infrastructure Network) player is Glow.

Qiao: Glow was part of the accelerator program we held in Portugal a few years ago. I remember they had a hard time raising funds after demo day.

Imran: Yes, but now they’ve created $20 million in annual recurring revenue (ARR). That’s an extraordinary transformation.

Trend Predictions: Stablecoins, Decentralized Infrastructure, and AI

Qiao: Shall we start discussing predictions?

Imran: Sure. Let’s talk about predictions for this year.

Qiao: My view is that the crypto industry is in a unique phase—it’s neither early enough that trends haven’t formed nor late enough that trends have started to decline. This means we can reference historical trends to speculate on what might happen in the next few years.

Qiao: For example, stablecoins will continue to grow. DePIN is also in a favorable position for growth. The total amount of on-chain activity may continue to increase, with a rising share compared to centralized exchange activity.

The chart you showed me this morning indicates that over the past few years, the market share of decentralized exchanges (DEXs) has grown from 10% to 30%, performing strongly compared to centralized exchanges. What else? AI and cryptocurrency? I mean, if we continue to extrapolate price trends, prices will rise. Now is the time for trends in the cryptocurrency space to extend.

Imran: From my perspective, before 2016, cryptocurrency had almost no real use cases. In 2017, there was a wave of experimental ideas—do you remember Mist Wallet? It was one of the earliest wallets for Ethereum. They tried to create a dashboard that could do everything but failed because the technology was too advanced. By 2020, we began exploring specific niche markets like DeFi and NFTs. And now, we are exploring all the areas that cryptocurrency can touch.

For instance, decentralized science (DeSci). Some people now think it resembles a scam, but over time, some legitimate projects will gradually emerge. The same goes for AI and cryptocurrency, as well as all other fields. Over the past decade, we’ve tested many areas one by one. I feel we are close to exploring most of the possibilities.

Qiao: For me, the two most obvious directions at the intersection of cryptocurrency and these fields are cross-industry financing and collaboration. Cross-industry financing has yet to be fully tapped—you can almost fund any industry with cryptocurrency. Collaboration is at the core of decentralized physical infrastructure networks (DePIN). This is not a single vertical but a way of operating across multiple industries. Essentially, DePIN is a method of coordinating capital and resources through tokens. It’s a means, not an end goal.

Imran: I agree, that might account for 80% of the value, but the remaining 20% still has some novel and interesting use cases worth looking forward to.

Qiao: There’s also fintech, especially stablecoins. This will be an important direction—we are essentially extrapolating based on historical trends.

Imran: But we haven’t seen any mainstream emerging banks (neo-banks) adopt stablecoins on a large scale.

Qiao: Like Nubank in Brazil.

Imran: I’m thinking of more emerging banks in the U.S. or English-speaking countries.

Qiao: I have indeed seen a lot of content about stablecoin adoption in the U.S., but I don’t think it will happen in the short term. Stablecoins are very useful for cross-border payments, hedging against local currency depreciation, micropayments, and eliminating chargeback risks for merchants. However, these use cases are not particularly relevant in the U.S. because payment systems like credit cards and Apple Pay have already taken dominance.

Imran: Absolutely right. In the U.S., the dollar is essentially already digital, and stablecoins are just an extension of it, providing little added value.

Qiao: But outside the U.S.—especially in developing countries—stablecoins have improved the traditional financial rails by tenfold. That’s also why most of the stablecoin startups we invest in are located in Africa, Southeast Asia, and South America.

Imran: From a global perspective, stablecoins or "crypto dollars" meet a huge demand outside the U.S. They can drive the formation of prediction markets and global capital pools, which are harder to achieve in traditional systems.

AI Social Tokens: AI Agents as New Social Influence Tools

Qiao: Perhaps the intersection of cryptocurrency and artificial intelligence will give rise to something new. I feel the most interesting products won’t brand themselves as "AI crypto." Instead, they will be products that make users feel great, and users might not even realize that AI or crypto technology is being used behind the scenes. That’s the distinction. The current AI agent hype is too overt—it mainly revolves around tokens and agents based on large language models (LLMs), concentrated on platforms like Twitter. It doesn’t seem like a truly great product; it feels more like a trend being forcefully pushed.

Imran: I understand your point, but I’ve noticed a certain "magic" with AIXBT. This magic lies in its ability to provide useful insights on platforms like Twitter. For example, its tweets often analyze various products, pointing out their growth potential, possible revenue scales, and other key metrics. Essentially, it’s replacing the role of key opinion leaders (KOLs).

What’s surprising is its dissemination ability—each tweet attracts about 150,000 views. When you think about having a token that can promote its own products, this concept becomes even more interesting. Essentially, it’s like an advertising network. For instance, Woody once proposed to AIXBT to use its avatar for a reward of $10,000 to $20,000 in Quantum Cats PFP. AIXBT accepted, and as a result, the price of Quantum Cats went up. Clearly, some significant changes are happening here.

Qiao: Yes, the idea of AI agents replacing KOLs has been around for a while. This trend started in Japan, where virtual characters have become very popular.

Imran: Yes, but the uniqueness here is that the community can influence the behavior of the agent through tokens. By purchasing tokens, users can theoretically shape the agent’s behavior patterns over time.

Qiao: So how does purchasing tokens allow users to influence the agent?

Imran: The core lies in creating a constrained terminal. Currently, I’m not entirely convinced by AIXBT’s existing token model. It requires users to purchase tokens to access its terminal interface, which is somewhat similar to ChatGPT’s subscription model, but here it’s achieved through token ownership. As the token price rises, the barrier to accessing the terminal may become increasingly high, which could limit widespread adoption among users.

However, I think this model has broader potential. If we view the agent as an application that has mastered end-user relationships, it blurs the lines between KOLs and bots. People’s interactions with AIXBT feel like conversations with a real person. If token holders can influence the agent’s behavior, such as promoting their products, that would be a disruptive innovation.

Governance and Controversies of AI Social Tokens

Qiao: So, the token essentially acts like a governance token?

Imran: Exactly. Through tokens, users can guide the agent’s behavior, making it a tool for influencing public opinion or promoting products. While this hasn’t been fully realized yet, it’s the direction I hope this concept will develop in the future.

Currently, AIXBT’s token is essentially a governance token. Users purchase tokens to access AIXBT’s terminal, which is similar to a ChatGPT-style product. However, I envision a more interactive model, like the one we are developing at Slop. In this model, token holders can influence the AI agent, prompting it to promote their products or ideas. Given the network effects and high visibility—like each tweet getting 150,000 views—this governance model could become an effective tool for capturing scarce mindshare.

Qiao: The entire hype around AI agents is actually centered around social tokens. We’ve been discussing social token experiments for years, but these tokens have never gained attention in "human" groups. One major reason is that most KOLs don’t want to be tied to tokens. AI agents, however, don’t care.

Imran: That’s true, and there’s another reason: holding tokens brings performance pressure. KOLs who hold tokens face enormous expectations from the community. Crypto users, especially early adopters, are often skeptical, and if those expectations aren’t met, they will quickly criticize.

Qiao: That makes sense. AI agents lack emotions or personal interests, so social tokens are more effective with them. Agents will simply execute tasks without resistance or concerns.

Imran: This aligns with one of our core hypotheses at Slop. The first point is AI agents governed by tokens. The second point is the area Sphere is exploring, which is converting natural language instructions into actionable trades.

Qiao: I agree with this perspective. The concept of converting natural language into executable results has existed for a while, but it remains a tricky problem that hasn’t been fully solved.

Trump's Policies, Bitcoin Predictions, and Market Cycle Shifts

Imran: Meanwhile, Trump is stirring up discussions again. He criticizes high interest rates, opposes new wind turbines, and even suggests renaming the Gulf of Mexico to "American Gulf." Clearly, his aim is to boost the market.

Qiao: Trump indeed wants to push the market higher.

Imran: The next four years look like a very favorable market for "hype."

Qiao: I think the peak of Bitcoin could be between $140,000 and $500,000, depending on macro conditions and the new government’s policies.

Imran: We’ve discussed this before—you said $500,000, I said $250,000. That’s a wide range, but it ultimately depends on the direction of macro developments.

Qiao: Additionally, the memecoin craze has been replaced by the AI agent hype. Memecoins are no longer the dominant force in the market.

Imran: Agreed. By the way, Murad is back, and his portfolio is performing exceptionally well. The last time I checked, his assets were valued at $53.5 million, with significant investments in Solana and other blockchains.

Qiao: Will we see a Solana ETF?

Imran: I think it’s very likely. Meanwhile, Trump is making bold promises, like significant tax cuts, raising wages, and improving income. Inflation is still high, and interest rates are a key issue. Let’s see what decision the Federal Reserve makes this month—whether to cut rates or maintain the current level.

Qiao: They just need a political excuse to lower interest rates.

Imran: Exactly. No one wants to take responsibility or damage their reputation. Everyone is overly cautious right now.

Qiao: I should go have lunch. I can’t wait for the bull market to end and prepare for the next cycle.

Portfolio Discussion: BTC, AI Agents, and the Decline of Memecoins

Imran: Okay, before you go, let’s talk about the portfolio. What does your current portfolio look like?

Qiao: I’m currently all in—BTC and AI agents.

Imran: That’s a solid barbell strategy. I also hold HyperLiquid. I think it’s a strong beta investment.

Qiao: HyperLiquid is indeed undervalued, but I don’t think it can outperform AI agents from now on.

Imran: I agree; it probably won’t outperform AI agents. However, the combination of BTC, HyperLiquid, and AI agents is indeed a good barbell strategy. What about memecoins? Do you think they will perform?

Qiao: No, I think the memecoin craze is over.

Imran: You say the craze is over, but you haven’t sold yet. It only counts as over when you actually sell.

Qiao: That’s true, but for me, it’s already over.

Imran: Well, let’s leave it at that.