The following text is organized from the series Twitter Space #Dialogue with DEV, hosted by FC, founding partner of SevenX Ventures, Twitter @FC_0X 0
Guest this episode: Marvin, founder of Phala Network, Twitter @marvin_tong.
First time as a Dev: No top VC endorsement, but got on Binance with community support.
After graduation, Marvin became an ordinary worker, working as a product manager in a large company. In 2018, the disruptive revolution brought by smart contracts made him and his brother @bgmshana, who was working in the Gmail team doing Machine Learning, come up with the idea of starting a business.
Which track to choose? Marvin and the team have a logic: the core value of Web3 is to solve the trust crisis, which will give birth to new innovative models, and the value of these innovations will be captured by the technology with the highest level of trust.
And TEE is just such a technology. Thus Phala was established, a decentralized cloud using TEE as hardware. After experiencing the initial 'hitting walls everywhere', entering the Polkadot ecosystem became a turning point for the project. After a period of building, Phala became one of the leading projects in the Polkadot ecosystem and seized the opportunity to successfully list on Huobi and Binance.
Regarding token listings, Marvin believes that one thing they did very right at the time was the fair launch of the token. Most of Phala's tokens are mined, and users do not need to run dedicated nodes; as long as they have TEE hardware, they can mine coins.
This has garnered widespread support from the community, which in turn has brought a lot of traffic. Although there is no endorsement from top VCs and no 'first-class' resume team, the eye-catching data metrics from the community and user dimensions, aka the grassroots support, coupled with the trend of Polkadot, led to the project's token listing.
But the trade-off of Fairlaunch still exists to this day. Because the team only holds 5% of the tokens, many capital collaborations sought after the listing have fallen through.
The transformation journey of Dev: go to the US, find PMF.
From 2021 to 2022, Defi Summer and NFT Summer were a very painful phase for Marvin and the team. They continually tried to find some use cases, but unfortunately, there were no results.
Marvin began to reflect; he had a vague feeling that in these years in crypto, due to some pre-listing, post-listing, and other game rules of crypto itself, he overlooked the points that truly need to be focused on in projects. He came to two conclusions:
1. Any Infra that really has users needs to find PMF.
2. The entire team must serve this matter before finding PMF, without being distracted by doing anything else.
"Of course, the premise of this conclusion is that the founders themselves need to understand what they want. We believe TEE still has a ten-thousand-fold potential that has not been tapped, so we want to pursue this. If we fail, then so be it. This direction. Focus on a direction and become the leader in that direction."
In order to find a breakthrough, Marvin and others moved to the US in 2023. One motivation was that doing infrastructure services for enterprises and developers could truly have users with PMF projects, possibly 70%-80% are in the US. Hence, they needed to be closer to this market, to know these people, to understand how they work. They talked to projects, participated in hackathons, and conversed with developers to find their problems, and since last year, they have had some PMF systems.
What does it feel like to find PMF?
1. Someone is really willing to pay for your things.
2. The way of use is very peculiar, usually not the way you expect.
It’s like selling bricks and asking everyone to build houses, but nobody buys. Then one day, suddenly someone wants to buy your bricks to put out the fire on their head; that’s roughly the feeling of finding PMF. Once this scenario appears, what needs to be done is to find the person whose hair is on fire and change the product form of the bricks to extinguish the fire faster.
Key points of transformation: how are important partner collaborators onboarded?
Throughout 2024, Phala will basically be onboarding customers. During this process, Marvin mentioned two key teams:
1. Flashbots
Flashbots is the key team that discovered the necessity of using TEE to solve the censorship problem off-chain on Ethereum. They also found Phala while trying to solve the TEE problem and had a series of technical exchanges and collaborations. Now Phala is also providing some services for Flashbots.
2. ELIZA
Marvin met Shawn in a hacker house and told him that if AI Agents could run inside TEE, they could achieve fully autonomous Agents, which is the 'End Game' of AI Agents in Crypto. Shawn was very buy-in to this logic, and after that, some of Phala's developers began to contribute more deeply to the ELIZA ecosystem.
Also because of this opportunity, Marvin launched spore.fun last December.
The story behind the launch of spore.
After getting to know Shawn and having in-depth discussions, Marvin had the idea of creating a case of ELIZA + TEE, with two routes:
1. Create Unrugable AI. After running the Agent inside TEE, it can eliminate human-made rugs. The product that Marvin and they collaborated on is called AI POOL, and they provide technical support.
2. Create Autonomous AI, to reflect the idea of complete autonomy to the extreme, leading to spore.
When is AI truly autonomous? It is when it does not need humans to support it. Currently, all AI on the market still require humans to pay the bills; developers are quietly paying Google, paying for the API of OpenAI's large models. What Marvin wants to do is to run the Agent inside TEE, then it pays itself, earns a living. If it makes a lot, let it replicate its genes, evolve, compete, and finally, survival of the fittest.
After the launch of spore, it received much love, with an ATH market value of possibly 80M, prompting Marvin to start taking this product seriously and operate and develop it as a separate project.
"In fact, after GPT 3.5, we found that more and more AI segment projects are willing to try our products, or at least willing to spend time listening to our Pitch. Later, as more and more AI clients came on board, we also began to focus on serving well in the AI track."
When Dev shifts to a trading perspective, how does he describe the potential of his project?
About Phala
First of all, if you believe that one day the penetration rate of Web3 can grow from today's possible 10% to 90%, then this process will definitely require low-cost, trustless technology to be compatible with Web2 software, because rewriting Web2 software is impossible. And the foundational infrastructure track needed for Web3 Mass Adoption, from the current perspective, Phala is in a leading position, and if not modest, it is number one.
Secondly, all problems stem from insufficient growth. Issues like insufficient chips for the dealer, not enough capital, etc., are insignificant in the face of explosive growth, so we should start to focus on the growth potential of the project.
In the end, quantitative data may not be very meaningful, including so-called value capture and profits, etc. What matters is the application scenario, that is, whether more people are using Phala's TEE, and who is using Phala's TEE.
Let’s talk about spore.
Now its value is brought by the team + topic, and this product is still being continuously built by a passionate team, so its growth potential is reflected in:
spore could become the most popular launchpad in the ai16z ecosystem, and this launchpad is also AI autonomously generated.
"Platforms like Virtual, which are excellent low-code Agent platforms, produce AI with an intelligence level of sixty to seventy. The AI generated on spore also has a similar intelligence level. The difference is that on Virtual, with resource stacking or a good team, there’s a 1% chance of elevating an AI from sixty or seventy to ninety, like AIXBT, but the remaining long-tail AIs don’t differ from those on spore."
Therefore, spore has a huge opportunity. Now many teams with ideas, or those that have previously developed good products but cannot issue tokens, can achieve this through spore.
The pioneer of AI Agents does not need any AI knowledge. It has unique skills added to the AI automatically generated agents rated sixty to seventy, turning it into something rated eighty to ninety.
2. All Agents produced on spore automatically run under a TEE ecological layer, using ELIZA's framework by default, with tokens issued on SOLANA, and are interactive.
What dimensions should a Dev consider?
Marvin describes the current Crypto market as extremely retail-oriented and extremely attention-driven.
He made an analogy with the development of live streaming sales. Initially, unknown internet celebrities and actors were broadcasting, and gradually more and more hosts came in. At that time, three types of people would emerge: 1) The audience, who finds it interesting and gets addicted every day, tipping and buying things; 2) The despiser, who criticizes from a moral standpoint, thinking this is sensationalism and a decline of societal standards; 3) The practitioner, who thinks, if in the future this world is all about selling things, what does everyone need? What do I need to do to capture market growth?
The current Crypto market is also the same; even the president has released coins. As a 'practitioner', one should think: if issuing coins becomes as common as air, something everyone does to express their needs, what should I become? A seller? A supplier for the sellers? A tool in the process? Or become the short video platform itself?
If there is no influencer system, it's not easy to become a seller.
Becoming a supplier, it’s best to find active items that are not yet widely circulated or scarce in the market, and then sell them through binding channels. Many teams are suitable for doing this, and many people are already doing it.
Providing tools, becoming a platform or part of a platform is a process of finding PMF, with a relatively low win rate. 'But like GMGN, it captures both demand and trends, possibly a product that comes out of a team and a way of doing things driven by extreme practice.'
In fact, for Dev, under the premise of being unable to predict the market long-term, there is no need to think about what the market will be like two years later. What needs to be clear is what you want and what you can do.
What should a Dev's Stop Doing List include?
"Don’t just blindly follow others. I think the thing I regret the most is not thinking from a first principles perspective to build my uniqueness."
The core logic of business is that you can provide value that others cannot. All actions, what product to make, what service to provide, ultimately come down to this point.
Once you slack off in deep thinking and first principles, the cost is that you will find that although you have done a lot of things and may have some effects, it is a waste of time and resources because it may not be what you really want.
Regardless of whether he has industry experience in crypto or not, he should quickly distinguish which are the rules of the game he needs to master to enter the industry, and which are the things he finds valuable through first principles thinking. Even if he's wrong, his speed of iterating through negative and positive feedback will be faster than others. If he doesn't realize this at the very beginning, it may lead to a lot of detours.
Trading opportunities in Crypto AI.
If Marvin brings in a trading perspective, his method of observing the market is not to look at mindshare, because that object is people, but to focus on the content itself, looking at the hottest Posts related to AI topics in Kaito. Reading the top 20 Posts every day is like looking at K-lines, finding a market sense, discovering what the top content, the most liked and most shared by big shots, is talking about regarding AI, and slowly getting to know the trends.
Regarding new directions for AI, Marvin focuses on two aspects:
1. AiFi
AiFi is not simply Ai + DeFi; it is not just an Agent that can swap or a DeFi that can adjust AAVE that constitutes AiFi. Marvin believes this has practical value but lacks imagination. AiFi is likely just an economic exchange capability of Agents.
2. swarm concept.
This is about how interactions between multiple AI Agents can hit interesting scenarios, rather than simply returning search results. So how to find it? Just as mentioned earlier, look at which posts are getting the most likes and shares daily in terms of mindshare; don’t look at people, look at posts, and slowly develop that feeling.
Finally, Marvin's little bells have three:
1. @NousResearch team's @SH L0 MS and @ropirito
Recommendation reason: Strong builder ability, the development capability and cultural precision of AI are excellent, and the things shaped have very high mindshare.
2. @shawmakesmagic
3. @yoheinakajima
Recommendation reason: A strong True Builder in AI, not only impressive in technology, but also very close to the market in terms of product intuition, able to shape what everyone wants.
FC Written at the End
From the perspective of results abstraction, Marvin, as a Dev, did a few things right:
1. Choose the right track with demand.
2. Know where the dominant players are and stay with them.
3. Because the right track was chosen, they encountered a new opportunity and met the key person in that opportunity.
4. Know the rules of crypto, how to play the secondary market, how to play the primary market, where resources are, and how to leverage them.
5. Luck and persistence.
Hope to bring some inspiration to everyone.
Thanks again to Marvin for participating in the dialogue with Dev. The Space audio will be updated gradually on Xiaoyuzhou, and we will separately open a podcast column. Any suggestions for the new series are also welcome to leave comments and interact.