Author: Wu Talks Blockchain

In this episode of Wu Talks Podcast, DINQ co-founder Sam Gao and Kelvin Sun conducted a systematic discussion around AI Agents, crypto infrastructure, and future organizational forms. The guests pointed out that the reason early AI Agent projects struggled to take off was fundamentally due to the immaturity of large model reasoning capabilities at that time. As model capabilities improve, agents are shifting from "answering questions" to "calling tools and executing tasks directly," which also revitalizes the integration of AI and crypto in practical applications.

When discussing the relationship between AI and crypto, both guests believe that most AI developers do not harbor hostility towards crypto but are relatively indifferent and lack a direct sense of connection. However, at a more fundamental level, crypto might still become an essential infrastructure in the AI era, particularly in three aspects: providing agents with a permissionless native payment system, enhancing the trustworthy verification of models and results through cryptographic tools like zero-knowledge proofs, and offering possibilities for a decentralized computing network in the future. Therefore, they conclude that AI and crypto are not likely to be in a replacement relationship in the long term, but instead are more likely to deeply complement each other in payment, verification, and infrastructure.

Regarding the entrepreneurial environment and industry culture, Sam and Kelvin further compared the differences between AI and crypto: the former allows small teams, individual capabilities, and work-oriented outputs to stand out more easily, while the latter is more heavily influenced by capital, narratives, and short-term incentives. They believe that AI is driving organizational forms to shift from traditional large teams to smaller teams that are lighter, more modular, and more flexible, even making some organizational visions previously envisioned by DAOs gradually executable in reality. From a longer-term perspective, the truly noteworthy opportunities may emerge in the intersection of AI Agents and crypto infrastructure.

The views expressed by the guests do not represent the positions of Wu Talks and do not constitute any investment advice; please strictly adhere to local laws and regulations.

The audio transcription is completed by GPT and may contain errors. For the complete podcast, please listen on Xiaoyuzhou, YT, etc.

From Civil Engineering to AI Entrepreneurship, to an Agent-Driven Talent Platform

Cat Brother: Welcome to "Wu Talks Non-Crypto." Our guests this episode are Sam Gao and Kelvin.

Sam does not have a computer science background but transitioned from civil engineering to AI, focusing on algorithm and artificial intelligence research, having worked for Alibaba's DAMO Academy for four years. After leaving in 2024, he began exploring the AI Agent direction, participating in the well-regarded work related to elizaOS last year, and is currently focusing more on AI-native product entrepreneurship.

Kelvin has long been deeply engaged in talent recruitment, talent identification, and organizational matching, with extensive observations of the talent market and entrepreneurial teams. Based on the complementarity of technology and talent directions, the two are currently jointly developing DINQ—a talent platform driven by agents, hoping to reconstruct talent discovery, identification, and matching using AI, which can be understood as LinkedIn in the era of Web4. Next, please each introduce your experiences and what you are currently working on.

Sam: I initially had a background in civil engineering and construction. During my master's program, I began systematically self-studying due to the rise of AI and later entered Alibaba's DAMO Academy through open-source contributions and writing articles, working within the Tongyi Lab system. After 2024, I mainly participated in some open-source projects, focusing on AI agents and coincidentally participated in the elizaOS technical white paper related work. By 2025, I started seriously thinking about the long-term value I could provide, and after meeting Kelvin, we began researching and starting a business around this direction.

Kelvin: The first half of my career was primarily in recruitment, typically entering related companies based on which industry was hottest. I initially did recruitment for foreign companies in China, then moved to real estate, and subsequently entered the capital field behind the internet. At Sequoia Capital, I was mainly responsible for recruiting talent in tech, TMT, and consumer sectors, particularly recruiting executives for invested companies. After leaving Sequoia, I started several entrepreneurial ventures and tried out HR Tech, cross-border e-commerce, and crypto directions. A few years ago, I connected with Sam through an industry veteran in crypto, and after some discussion, we confirmed this current direction and have continued to this day.

Starting to Create Tracking Products Because On-chain Data Tools Are Too Expensive

Cat Brother: Let's start with Sam's work background. You mentioned earlier that you initially focused on AI; how did you come into contact with AI, and how did you get into the crypto industry? Why did you later decide to work on projects like Eliza Labs and elizaOS, which combine AI and crypto?

Sam: I began to engage with crypto around late 2021 to early 2022. Friends around me were involved in related projects, and at first, I was just interested in understanding it, later realizing I preferred data-related directions.

At that time, typical products were Nansen and Dune, especially Nansen. It was expensive, yet many users were willing to pay, which made me realize that on-chain data services are highly valuable. So I sought out a friend to jointly create a tracking product similar to Nansen during our spare time, mainly focused on "whale tracking" on Ethereum and BTC.

Later, I continued to work on AI while keeping an eye on opportunities where AI could intersect with crypto. By the second half of 2024, when AI agents began to really gain traction in the crypto space, I felt it was a very natural entry point. Plus, the Eliza team happened to need someone knowledgeable in algorithms, which I could provide, so I got involved.

Kelvin: This wave of AI was not something people actively chose to embrace; it was something everyone had to enter. It is no longer an independent industry but more like a new foundational environment. For everyone, the key is figuring out what the relationship between AI and oneself is and finding a logically coherent position.

For me, talent and recruitment have always been what I excel at and invested the most time in, which naturally leads me to think about how AI could integrate with recruitment. At that moment, I was pondering this question, leading me to talk with Sam.

Sam is himself an AI talent and has many of his own views on AI applications in recruitment. One of the discussions we had most was about what fundamental changes AI would bring to this traditional industry.

Additionally, I also consider myself among the early users of AI, having used everything from GPT-3.5 to early Midjourney, so I naturally felt this was a suitable opportunity for me to enter. Hence, we began to pursue this endeavor.

From On-chain Data Tools to elizaOS: Project Ideas and Controversy Insights

Cat Brother: Nansen was at its peak around 2021, and you mentioned it raised $80 million in its Series A. At that time, a significant part of its business was helping Ethereum L2 and cross-chain bridge projects identify airdrop participants, screening out users who were merely trying to profit without contributing, and accordingly establishing distribution rules.

Another project you later participated in was elizaOS, and you seem to have been one of the authors of its white paper. This project was extremely popular, with a market cap once peaking at about $100 million, making it one of the most highly regarded AI Agent projects at that time. However, its token price later fell significantly, with the controversy surrounding the uppercase and lowercase "Eliza" tokens causing community division and debate.

I'd like you to share from an internal perspective how the Eliza project started and gained recognition, why the uppercase/lowercase争议 arose, and why it ultimately couldn't continue, leading even core developers like you to choose to leave.

Sam: Firstly, I was only there to help them with some algorithm-related work. Secondly, the token behind this project, I believe is called AI16Z, not the uppercase or lowercase Eliza. The two "Eliza" projects you mentioned are actually scams and unrelated to the core framework. The framework itself was essentially to serve the AI16Z token, that's the situation.

Cat Brother: My earlier understanding was that AI16Z was somewhat mimicking a16z, trying to create a meme-style VC; while elizaOS was attempting to create an ecosystem for AI Agents. I had interpreted it that way. It seems my understanding might not be accurate. It's actually correct to say that your white paper mainly serves the AI16Z token, right?

Sam: The initial problem the project aimed to solve was quite straightforward: everyone believed AI Agents would be the future, but crypto users held many fragmented assets across chains, while cross-chain bridges, lending, arbitrage, and storage protocols were too complex for the average person to effectively use. Thus, the original idea was whether an agent could automatically help users complete these on-chain operations.

Later, the founder Shaw expanded the direction further, seeking to create not just trading agents but a more general AI Agent capable of not only handling on-chain operations but also retrieving news, playing games, and even participating in game creation and economic systems comprehension. Essentially, it aimed to create a general intelligent assistant capable of executing various tasks.

However, I viewed the major immaturity at that time as the lack of critical reasoning abilities in large models, which was a core limitation preventing early AI Agents from truly becoming viable. Even so, elizaOS was still quite forward-thinking, both in terms of its technological direction and language choices.

For instance, most agent frameworks predominantly used Python at that time, but considering the crypto world where much of the infrastructure was already based on TypeScript, JavaScript, and Rust, continuing to use Python was not only possibly inadequate in performance but also complicated for integrating on-chain atomic operations. Hence, we ultimately chose to use TypeScript as the underlying language, which was already a relatively new direction then.

Moreover, elizaOS was also among the early projects to emphasize the "personality" of agents, not just enabling them to complete tasks but aiming for them to possess personality traits, thereby extending playful interactions. Coupled with the project fostering discussions around secure execution environments like FHE and TEE at that time, it was indeed an early yet quite avant-garde attempt.

Cat Brother: You mentioned that when you made elizaOS, after the project gained popularity, it indeed drove interest in the overall AI Agent arena. Based on what you've said, you feel that one main reason it couldn't sustain itself later was that the critical reasoning capabilities of large models were still inadequate at that time, is that right?

Sam: Yes, to be more accurate, there was indeed no genuinely mature reasoning capability then. The reasoning capacity of large models only became widely recognized after OpenAI announced related directions in September 2024, with the actual model releases occurring in December. At that point, most people still did not fully understand this.

Cat Brother: The market was bustling, and many people indeed didn’t thoroughly research this direction. Later, after DeepSeek emerged, the team gradually became less cohesive, and fewer people continued to push forward. Looking at it now, with rapid improvements in the reasoning capabilities of large models and accelerated model iterations, if today's projects were to restart or continue the path of elizaOS, do you think there is still development space for them? Or have they actually become outdated?

Sam: I believe the technology itself is certainly promising, and it is natural that AI will take over more and more tasks in the future. However, creating a truly explosive project often relies not only on capability but also on timing and luck, and it's tough to ensure success purely through planning.

So if we were to restart such projects now, I think the key would still hinge on the market environment and specific timing, rather than a strictly orderly approach guaranteeing the creation of a phenomenon-level product. This endeavor itself is indeed very difficult.

Viewing elizaOS from an Observer's Perspective: The Personal Drive Behind the Hype Project

Cat Brother: Do you still keep in touch with Shaw now? Do you still exchange technical insights? Furthermore, regarding the previous controversy over the uppercase and lowercase "Eliza" tokens, did you observe it closely at the time? I recall there seemed to have been an initial version released, then Shaw stated that it was not the official version but one deployed by the community, leading to a new official version being supported, which eventually spurred community division and price fluctuations. How did this occur? Did Shaw communicate with you about this?

Sam: No, we were primarily focused on technical matters.

Kelvin: I can’t really answer the second half because at that time, Eliza appeared to be quite a mysterious organization. Unless you knew exactly who was doing what internally, it was hard for outsiders to see too many details. Take the White Paper, for instance; if I hadn’t later found out Sam was involved, I wouldn’t have known beforehand. To outsiders, the only person people really recognized was Shaw.

Yet the project was indeed extremely popular, and this was acknowledged in the market. Back then, while I was at Galaxy, we noticed whatever track was hot, so we were certainly aware of it. However, as an observer, we felt more how popular it was rather than seeing too much internal information.

Later, as luck would have it, I got to know Sam through personal connections. To some extent, the fame of Eliza itself proved the capabilities and backgrounds of the participants, requiring little additional explanation.

If I had to share a broader feeling, I would say that whether it's Crypto AI or the current AI industry, there's a strong tendency towards personal heroism. Often, it is a few individuals who boldly innovate and bring things into being; whereas the previous generation of internet entrepreneurship resembled collective efforts where typically only the boss was known, with a whole group behind actually driving things forward.

AI Developers Are Mostly Indifferent to Crypto, but Agent-Native Payments May Still Grow on the Chain

Cat Brother: Recently, we've also mentioned OpenClaw. Its founder seems to have a distaste for crypto, as after the project's renaming, there was a period when a meme coin rushed to garner attention—issuing tokens ahead and even registering the related domain names. He later tweeted, clearly stating a desire to distance himself from crypto.

From your observations, is it the case that not just the OpenClaw founder, but many AI developers are indeed averse to the crypto industry? Overall, how does the AI developer community currently view crypto?

Sam: I think generally speaking, most AI developers are indifferent to crypto, treating it more like a parallel track. The aversions you mentioned are more residual effects of meme fervor—there's always someone rushing to issue tokens or grab traffic during a hot topic, which can indeed annoy some AI practitioners.

However, on a broader scale, most people do not have a strong repulsion towards the entire crypto industry. Because what everyone truly values is utility: whoever can solve a problem is worth something.

If AI circles really expressed widespread aversion to crypto, then a product like OpenRouter would likely struggle to exist. Its founder has a strong crypto background, yet that didn’t prevent it from becoming one of the key transition platforms in the large model field, indicating that the market is not naturally resistant to teams with crypto backgrounds.

Thus, I do not believe that crypto has a negative impact on AI. On the contrary, I think its greatest value lies in providing agents with native payment capabilities. Currently, agents still lack a genuine consumer and payment system, and once this issue is resolved, many product forms and business models will undergo significant changes.

Because of this, I pay particular attention to directions like the x402 that Coinbase is promoting. In my view, if a native payment system suitable for agents is to emerge in the future, it will likely not be built on the traditional fiat system.

Cat Brother: When you mentioned "indifferent," are you saying that AI developers do not harbor obvious animosity towards crypto, nor do they find it particularly attractive, instead holding a relatively neutral stance? Or is it closer to a sense of "not really caring," feeling that this matter has little connection to them? Which of these interpretations aligns more closely with what you’re saying?

Sam: Closer to the latter, but not a negative "not care." What I mean is more like: I know there’s something like this, but it develops according to its logic and doesn’t have a direct relationship with me. It's not rejection or complete disregard, but rather a relatively aloof, observational stance.

Cat Brother: You mentioned the OpenSea CTO, which reminded me of a rumor. OpenAI founder Sam Altman also has Worldcoin, right? There have been claims that Worldcoin, to some extent, was meant to finance OpenAI or related AI endeavors because early AI projects had difficulty raising funds, while crypto projects were rather easier to attract capital attention. Have you heard this claim? How truthful do you think it is?

Sam: There is some truth to this claim. Particularly in 2018 and 2019, OpenAI faced quite a difficult situation, undergoing changes both internally and externally, so they did indeed consider financing through ICOs and other crypto methods to relieve fundraising pressures.

Cat Brother: In other words, OpenAI, to some degree, almost brushed past crypto.

Sam: You could understand it that way. However, strictly speaking, that matter was not directly related to ChatGPT at that time. The real widespread recognition of the GPT path came after GPT-3, especially after 2022.

Will AI Practitioners Buy Tokens, and Is the Crypto Industry in Its "Bull Lady" Phase Again?

Cat Brother: From your observations, do the AI developers you encounter hold crypto assets like BTC or ETH? Do you have any positions yourself?

Sam: From the AI people I know, most have some holdings, mainly in mainstream assets like Bitcoin, Ethereum, and Solana. Most do not particularly focus on it but do have some configuration; of course, there are a few who are more interested in new things and even research how to participate in products like Polymarket.

Overall, I think this issue largely depends on personal interest and energy and does not have a particularly strong direct relationship with AI; the correlation between the two is not that high.

Cat Brother: Now let’s look at Kelvin's side from a statistical standpoint. You have worked at a crypto company, invested at Sequoia, and now are involved in AI-related talents and recruitment. What attitude do you observe among AI entrepreneurs and developers towards crypto? Do they hold crypto assets?

Kelvin: I think my observations are similar to Sam's. In the investment circle, crypto is no longer new; many VCs have been involved and experimenting for a long time, experiencing several cycles, making profits, and also incurring losses.

Also, talent is inherently fluid, and many people frequently shift between traditional industries, the internet, and the crypto industry. Where opportunities and funding exist, that’s where talent will flow.

Cat Brother: From the perspective of talent and funding, do you think crypto has now entered a relatively neglected phase again? After all, Bitcoin has also dropped quite a bit recently.

Kelvin: If viewed from the perspective of trend followers, there is indeed that feeling; but for long-term participants, this kind of fluctuation is not new, and they still expect the next cycle.

Thus, I've always felt that this time may not be a bad moment to enter the industry. Here, "entering" doesn’t just mean buying assets, but also includes working for related companies, starting a business, or directly engaging in the industry. Because you may just catch the next opportunity.

AI's Personal Heroism and the Structural Dilemmas in the Crypto Industry

Cat Brother: Sam, I’ve listened to a past podcast of yours from 2025. You mentioned that many AI developers, even those with strong technical skills, do not necessarily come from prestigious schools nor have substantial funding support, yet they can still produce excellent products. Behind this is a strong sense of "personal heroism," which Kelvin also pointed out.

In contrast, many projects in the crypto industry seem increasingly VC-led, making it harder for individual developers to come forward. Do you believe this structural difference somewhat suppresses innovation in crypto? Especially since from DeFi Summer till now, there seems to have been a long absence of almost universally recognized real innovations in the industry—could this be related?

Sam: I think there is indeed a connection. In the AI industry, many key decision-makers in projects are very young because they prove themselves through their work and results.

Conversely, the issue in the crypto industry is that once grassroots developers create something, funding quickly floods in, causing projects and individuals to spiral out of control. Many teams don’t have sufficiently clear directions or organizational capabilities, making it easy for them to become scattered. On the other hand, VCs often pursue quick returns, opting to "assemble projects" rather than slowly identify truly potential individuals. However, the teams and projects produced this way typically lack genuine competitiveness, as the results have shown.

Therefore, I believe the industry greatly needs research-oriented organizations to focus on long-term foundational technology and innovation exploration, such as how blockchain infrastructure can improve, or how agents can interact with oracles, zero-knowledge proofs, and reinforcement learning.

The problem is, the market is currently too impatient for quick results, making it hard for such people and organizations to survive. But if the industry continues to lack this long-term, technology-oriented force, it will be difficult to generate genuine innovation and attract new increments.

Cat Brother: Over the past two to three years, Vitalik has been researching foundational technologies like zero-knowledge proofs, but many criticize that these concepts have yet to materialize. In contrast, AI seems to integrate more easily with real-world scenarios and allows for more visible applications. Could it be that the difficulty of applying crypto technologies has inversely stifled research and innovation within the industry?

Sam: I believe there’s no direct relationship between them. Many great technologies initially seemed to have little real-world value. In AI's early days, whether it was playing small games or doing next-word predictions, those activities were not termed "useful"; they developed step by step.

Thus, the problem is not that what Vitalik is studying lacks value, but rather that the market is too short-sighted and lacks patience. Vitalik can propose direction, but what’s truly missing is a research team willing to execute and delve deeply over the long term. There are too few such people in the industry now, and many ideas end up remaining at the conceptual level.

Conversely, I feel that with the arrival of the AI era, there will be more people entering a state where "everyone is a researcher," actively studying models, data, and capability boundaries. Similarly, as long as environmental conditions allow, more and more people will be willing to research blockchain itself in the long term.

So the fundamental issue is not that the technology lacks value but rather that the market is too impatient to accommodate such long-term innovations.

Cat Brother: You are currently building a venture in the AI field, whereas before at Eliza you focused more on core technologies without being at the forefront. From your personal experience, what do you think is the biggest difference between working in the crypto and AI industries? Especially in terms of industry culture, working pace, and business models, what notably differs?

Sam: In my previous AI work, I was more within a corporate system, where there was relatively less direct pressure regarding results and survival. But crypto is different; it really tests a person's quick response, deep thinking, and execution capacity—it's a high-risk, high-reward industry.

By contrast, the AI industry is generally friendlier to most people; as long as you are smart enough and put in enough effort, the lower limit typically isn't too low.

However, looking at the historical development of technology, many genuinely important breakthroughs may not be the most recognized directions at present. So I have always felt that when the path you choose is not entirely aligned with the mainstream, you should not rush to doubt yourself but rather persist. Many significant opportunities often lie hidden in these non-mainstream directions.

Thus, my judgment remains simple: in the AI era, opportunities always exist; the key is whether you can continue to pursue them.

Cat Brother: In your experience, what clear differences are there in the entrepreneurial environments of AI and crypto? Do these differences reflect in what kinds of people are more easily recognized, supported, and amplified?

Kelvin: I believe the most significant difference is in the incentive mechanisms. Feedback and incentives in crypto tend to arrive faster but are also more short-lived; AI is relatively more balanced—the outcomes in innovation, whether in entrepreneurship or within institutions, can typically be seen within a year or two.

Nevertheless, I do not think the differences between the two are exceptionally large as both belong to industries that strongly encourage innovation. Both require new ideas and people and emphasize breaking old norms; sometimes, what needs breaking isn’t just from the former era but also what existed just last month.

From a talent perspective, there is significant commonality in both realms. Many key figures in current AI companies have actually achieved successes in the crypto industry in past years and have brought some mature and effective strategies from crypto to AI, which has benefited the development of the AI sector.

After Exchanges Enter the Game, What Kind of AI Agent Does Crypto Truly Need?

Cat Brother: Sam, recently many trading platforms, such as OKX, Binance, and Bitget, have introduced functions or products related to agents. I’m not sure if you have been following this. From your perspective, what stage has the entire AI Agent domain reached now? And what kind of agent products do you think the crypto industry truly needs?

Sam: In the early stages, many AI Agents were essentially informational products—user specifies needs, and they provide answers; however, this model primarily solved end problems, and the practical value was limited.

The current transformation is that platforms have begun to offer capabilities like skills, MCP, SDKs, and APIs, allowing agents to actively call tools and accomplish more complex task chains rather than just providing answers. Additionally, many agent systems now incorporate features like scheduled execution and continuous operation, increasing their proactivity.

I think future valuable AI Agents will likely need three characteristics.

First, they should understand users sufficiently. They should be able to analyze based on historical data, trading records, and behavioral preferences to actively execute operations directly rather than just staying at the "advising" level.

Second, they must possess genuine execution capability. In facing the complexity of multi-chain and multi-protocol environments, the value of agents lies not in telling you what to do, but in directly assisting with asset allocation and yield management, truly solving complex problems.

Third, they must ensure high security. Such agents will require more sophisticated security mechanisms in the future, such as sandbox environments, permission control, blocking abnormal transactions, and emergency switches to prevent loss of control or mistakes.

Thus, in summary, truly valuable agents not only analyze but also execute and must be safe and controllable.

Cat Brother: AI is now virtually omnipresent. From your perspectives, are there key aspects where AI would struggle to operate independently without crypto support? In other words, does AI have inherent dependencies on crypto technologies or assets in certain areas?

Sam: I believe there are mainly three directions.

First is payment. Agents currently do not have their own native payment system, and existing banks, credit cards, and payment platforms are fundamentally designed for humans. In contrast, crypto offers a permissionless account system, making it more plausible for agents to become true independent actors capable of making payments.

The second is trustworthy verification. For instance, when users invoke a certain model, how can they prove the platform hasn’t secretly swapped in a weaker version? These types of concerns usually require the assistance of cryptographic technologies like zero-knowledge proofs to verify model identities and the authenticity of results while protecting privacy.

The third is computing power and infrastructure. If a usable decentralized computing network were genuinely to emerge in the future, it could provide AI with lower-cost, more flexible training and research resources.

Thus, in my view, payment, verification, and computing power are three areas where AI can hardly separate itself from crypto.

If DINQ Gets on Track, I’d Prefer to Focus on Incubating and Supporting Young Talent

Cat Brother: Suppose you and Kelvin’s AI talent recruitment platform successfully gets on track and you no longer need to work personally, with AI managing everything for you. Would you consider, combining your background in AI, starting a new venture in the crypto industry? Regarding the three directions you mentioned earlier, do you see potential for that? Or do you personally have any willingness to pursue this?

Sam: From my personal perspective, the most core purpose of creating this platform is to help more talented young people stand out.

If DINQ truly gets on track, I would prefer to engage in offline activities, or build an incubator, or establish early-stage investment institutions. Something similar to AllianceDAO, aiming to support and aid young people, enabling them to accomplish greater things.

I believe this is what I genuinely want to do. As for whether there could be future combinations with crypto in terms of funding mechanisms, I am relatively open to that as well.

What are the Major Differences between Working in Crypto and AI?

Cat Brother: You are currently building a venture in the AI field but previously worked on algorithms and core technologies at Eliza. From your experience, what are the most noticeable differences in working across these two industries, particularly concerning industry culture, work pace, and business models?

Sam: My prior work in AI involved being within a company structure where I didn't have to directly bear much survival pressure; however, crypto is more brutal and places greater emphasis on an individual’s quick response, deep thinking, and execution capacity—it's a high-risk, high-reward industry.

In comparison, the AI industry tends to be kinder to most people; as long as someone is clever and hardworking enough, their baselines typically aren’t too low. However, I still believe that many genuinely crucial technological directions may not always be the most recognized ones currently.

So if your path diverges from the mainstream, you don’t need to second-guess yourself too early. Opportunities in the AI era always exist; what matters is whether you can continue to follow through.

Kelvin: In my view, the organizational style in crypto resembles traditional internet models, generally requiring more people to collaborate and leaning towards team-based operations.

However, AI, especially in its application layer, often has very lean teams where a few top talents effectively collaborate to get things done. This distinction is not merely a difference between AI and crypto but is more about the divergences in organizational methods between AI and the majority of past industries.

AI has also allowed many functions that previously depended on large corporations to begin functioning independently. Teams that once needed to reside within large organizations can now utilize AI tools to fulfill complete work chains, transforming into more flexible, modular, and pluggable small teams or external partners.

Therefore, I believe that one essential change brought about by AI is that organizations will become lighter and more agile.

One-Person Company, DAO, and AI's Reshaping of Crypto Organizational Forms

Cat Brother: Recently, a very popular concept called "one-person company" has emerged. A single person can run an entire company through various AI agents and assistants. Do you perceive this more as a gimmick, or do you think that as AI technology develops, it could genuinely become a reality, even becoming a norm?

Kelvin: I can only share my perspective. I think this notion is somewhat extreme. Certain individuals can genuinely accomplish many tasks alone, but this is not an entirely new concept. There have been many past examples where someone alone could illustrate their talents through platforms, essentially mirroring the previous generation's "one-person company" model.

However, I believe what’s more common is still small teams—like two to three people, three to five, or ten individuals. Many may serve as CEOs of their own companies while also functioning as a CTO in another or doing consulting for others. In other words, people will become more interconnected and collaboratively flexible, enabling quick collaborations and timely deliveries, commitments, and payments. I think this pattern will become increasingly prevalent.

Consequently, any company actually has the capacity to leverage the advantages of talent, as many individuals already offer their services in this small organizational form. This is a trend I've observed quite distinctly.

Cat Brother: It feels like this part of our conversation mirrors a popular concept from the crypto industry a few years ago regarding various DAOs, decentralized organizations. It seems that kind of state is returning.

Kelvin: I think it genuinely is. Why am I particularly sensitive to this? Because the "O" in DAO stands for Organization, if I remember correctly, and the subject of "organization" happens to be my area of expertise. I have always paid attention to how organizations are formed.

Thus, I was quite bought into the concept of DAOs back then. I believed that organizations composed of intelligent individuals should be like that: highly self-driven and not requiring everyone to sit together at work. Having a company with 20 people from 20 countries seems very reasonable to me. At least many of the outstanding organizations I’ve observed in the past have operated this way.

Cat Brother: So, could it be that the entire crypto industry benefits from the development of AI technologies? Many concepts that were previously envisioned but not realized may come to fruition thanks to AI. Extending this further, even though the AI industry may shock the crypto industry in terms of funding and talent in the short term, could it potentially assist the crypto industry in the long run by enhancing efficiency? Do you agree with this perspective?

Kelvin: I absolutely agree. Recently, I have interacted with some entrepreneurs in the crypto sector. At least from a narrative standpoint, many popular narratives in AI today can still be traced back to historical concepts from almost every crypto narrative in the past.

In other words, at least with respect to narrative innovation, crypto has consistently been ahead. You can fully extract many elements from it and recreate them using AI because the major difference between now and the past is that these ideas can genuinely be implemented; they are no longer stuck at the conceptual or narrative levels.

Including the project we are currently working on, I actually came across a company called Braintrust back in 2017, which aimed to create a "decentralized LinkedIn." Although its implementation then was different from today, I was very bought into that concept. Looking back now, I wonder if we have the chance today to realize the future that company envisioned back then.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。