Justin Sun, founder of TRON, used a long-form podcast conversation with Alan, the creator of the “thin-muscle theory,” to explain how he sees personal reinvention in the AI era. The discussion, compiled by TechFlowPost, moved across several topics: AI-managed daily workflows, low-cost self-upgrade strategies, global opportunity scouting, proactive health management, and Sun’s view of crypto as infrastructure for a machine economy.
Building a daily system around local files and AI agents
Asked about his recent training routine and current physical condition, Sun said he is 172 cm tall, weighs about 75 kg, and has a body fat ratio of roughly 29%. He said he plans to complete what he described as a full physique restructuring over the next three to six months.
Sun said much of his daily life is now handled through what he called full AI management. For meals, he takes a photo before eating and sends it to Claude, then sends another photo after finishing. Claude calculates calorie and protein intake and suggests an order for eating, such as prioritizing high-protein foods. He said he currently supplements with 25 grams of protein powder and 5 grams of creatine per day, with a daily protein target of 140 to 150 grams.
For training and sleep, Sun said he uses an Oura Ring to capture sleep and energy expenditure data and feeds that information directly into AI tools. His coach records workout videos, which are also uploaded for AI processing. The system identifies whether he is doing deadlifts or squats, logs the number of sets and the weight used, and plans the next increase under a progressive overload approach. Sun added that AI compiles his sleep, diet, and training burn each day into a visual personal health report, including figures such as 275 calories burned in a workout.
On the technical side, he said the key principles are data localization and model decoupling. Training records, photos, and daily analyses are stored as local files, especially Markdown (.md) documents. He said he set up dedicated projects so the AI writes results directly into those local .md files.
That design, he said, is meant to avoid dependence on any single vendor. If Claude works best today, he can call Claude’s API. If OpenAI’s coding interface or Gemini becomes stronger tomorrow, he can switch APIs and let a new AI agent read the historical .md files and continue the workflow without rebuilding the whole system. Sun said this sharply reduces the cost of migrating personal digital assets and changing software tools.
He extended the same logic to broader personal data management. In his view, traditional SaaS products trap health and work data inside isolated silos, leaving users exposed if a platform shuts down or raises prices. A local Markdown-plus-API structure, he said, gives individuals direct ownership of data while making it easier to tap whichever leading model is available.
What he would do at age 20 with less than RMB 1,000 a month
Alan then asked Sun how he would start over at age 20 with no resources, no network, and less than RMB 1,000 in monthly disposable income. Sun’s answer was that the current era is unusually favorable because AI has pushed the cost of accessing high-level knowledge down to almost nothing.
Even if a person could not afford a $200 monthly subscription, he said, they could still build an AI-based personal system using free tools. He mentioned domestic Chinese AI tools or APIs such as Zhipu and Kimi, as well as B.AI, which he described as a setup combining multiple large models. A regular smartphone, he said, would be enough. The first step would be to create folders on the phone and record one’s current situation, sleep, diet, and learning progress in documents, then ask free AI tools to diagnose the situation and suggest a path forward.
Sun said the next step is to install three operating systems:
- “Sun studies”: cognitive upgrading and a reworked understanding of what financial freedom actually means.
- “Thin-muscle”: treating physical health as a core asset and keeping the “carbon-based server” stable.
- “Yellow hair theory”: strong execution and the ability to block out external noise.
He said many people fail to get results not because information is unavailable, but because they cannot get through social resistance. In his telling, someone with higher-level thinking and better physical condition can easily become a target of doubt in the Chinese-speaking world. The “yellow hair theory,” as he framed it, acts as mental armor: trust science, trust data, trust AI-generated reasoning, and keep going even when outside voices turn hostile.
Sun’s suggested process is to spend two to four months accumulating personal data, then feed that data into several top models, including Claude, OpenAI, and Gemini through APIs. He said users can ask for upper, middle, and lower strategies, choose the middle option or the best one, and then keep tuning based on AI feedback. Pulling back from what he described as the hidden costs of social life and preserving energy for AI-assisted execution can widen the gap in understanding quickly, he argued.
“Map vision” in the AI era and the 37% benchmark model
When asked what young people should go all-in on over the next few years, Sun did not point to a single sector. Instead, he said most industries will be reshaped by AI, while human lifespan and time remain scarce. Opportunity is everywhere, he said. The real question is whether people know how to use AI to gain full map vision.
He cited an academic example to illustrate the point. Sun referred to a New York University professor working on the Navier-Stokes equations, one of the Millennium Prize problems. After the professor shared a core idea with OpenAI, Sun said, OpenAI brought in $7 million in compute and thousands of agent nodes, then followed that line of thought and solved the derivation in two days. For Sun, the point is that AI is breaking down old barriers in academic and industry circles, giving younger researchers a way to leap over much of the traditional institutional path.
From there he moved to a broader claim: if human knowledge, mathematical proofs, or historical culture are not formally verified and uploaded to the internet as AI training material, then in the next era that civilization may as well have never happened.
Applied to commercial decisions, Sun’s advice was to scout before committing. He compared the process to real-time strategy and MOBA games, where fast units are sent out to reveal the map. In practical terms, he said young people can use AI to model employment and industry prospects for 100 different majors over the next 10 years, then decide only after they can see the broader distribution of risks and resources.
He also referred to what he called a “37% benchmark decision model,” described in the conversation as the “100 houses model.” If a person is choosing a major or buying a home from a set of 100 options, he said, the first three or the first 30% should not be used for a final decision. They should establish a benchmark. Starting from the next set of options, once a candidate appears that beats the earlier benchmark on combined metrics, the person should act. In his view, people often fail either because they choose the first option too quickly or because they review every option and become paralyzed.
On business models, Sun argued for global arbitrage and mobility. He gave the example of regional price differences in AI service APIs, saying the same service can be priced very differently in markets such as Argentina, Turkey, Nigeria, and the United States. That gap, he said, can create legal business opportunities in token transfer hubs or cross-border demand matching. His framing of the market was broad: not just the current global population of 8 billion people, but also billions of future silicon-based agents.
That logic extends to personal financial choices. Sun said he strongly advises young people not to buy a house, buy a car, or get married before age 30 if doing so locks up their liquidity too early. Fixed assets and local cash flow obligations can tie someone to one place, he said, just when a major opportunity opens elsewhere, whether in Silicon Valley or an emerging market. In his view, keeping a light asset base and high mobility makes it easier to shift both the body and capital toward areas with higher returns on productivity.
Proactive health management and the role of muscle
Health management was another central theme. Alan asked why Sun puts muscle and resistance training in the category of the highest-return investment. Sun said he recently developed, through discussions with AI, a quantitative 100-point longevity model:
- Cardiopulmonary function: 25 points
- Muscle mass: 25 points
- Organ maintenance, anti-aging, genetics, and related factors: the remaining 50 points
Sun argued that scientific resistance training, including compound movements such as squats and deadlifts, can help a person capture the first 50 points at once by improving both cardiopulmonary condition and muscle. By contrast, he said, pure aerobic exercise such as running may improve the heart and lungs while also reducing muscle mass.
He described muscle as the body’s “U.S. dollar cash balance.” In periods of major illness, stress, or physiological shock, he said, the body will consume muscle first to repair mucosal tissue and organs. If the body has too little “muscle savings,” the whole carbon-based system becomes more fragile. He also called muscle a passive fat-burning shield, saying it continues to consume energy during sleep, improves insulin sensitivity, and helps prevent more than 90% of cardiovascular and cerebrovascular disease linked to obesity and metabolism.
Alan also asked about public figures such as Elon Musk, Warren Buffett, and Donald Trump, who are often cited in discussions about not needing fitness, and about the popularity of semaglutide-based weight-loss drugs such as Ozempic. Sun said using people like Musk, Buffett, and Trump as reasons to avoid resistance training is an ego trap in which one’s position shapes one’s worldview. In his telling, successful people may defend unhealthy habits because those habits fit their own state. But if the question is whether one wants to walk into a restaurant at 80 or be pushed in by wheelchair, the answer is obvious.
On semaglutide, Sun argued that the drug can reduce fat but also cause a large drop in muscle mass. That, he said, lowers basal metabolism sharply and creates a destructive chain: more drug use, lower metabolism, faster rebound after stopping, less muscle, and eventually frailty, fractures, and bedridden old age.
As for implementation, Sun said the scientific approach is proactive data-based screening. He said that under AI guidance, he went to a hospital for deep testing despite having no symptoms, monitoring indicators such as high-density and low-density lipoproteins and blood glucose. He recalled that a doctor, after seeing the testing list, asked whether he had used AI and said Sun was the first patient the doctor had seen who came in with no symptoms and was driven purely by AI-based health screening and data monitoring. For Sun, that captures the difference between proactive health management and reactive emergency care.
Crypto as programmable money for machines
The crypto section of the discussion centered on a basic question from Alan: what does cryptocurrency solve at the root level, and what role does TRON play in that system? Sun’s answer was direct. Crypto, he said, translates humanity’s old form of money into a programmable language that AI and machines can understand.
He said the current SWIFT settlement system traces its roots to the 1950s and to the post-World War II Allied telegraph structure. Even though the front ends of banks and firms such as China Merchants Bank and PayPal have been digitized, Sun said their back-end ledgers remain closed and centralized databases, which he believes are not compatible with silicon-based AI systems.
In Sun’s view, the AI era will involve thousands of autonomous agents running on their own and calling each other’s services. An agent may call another agent’s API or settle charges based on millisecond-level compute usage, and the amount involved may be only a fraction of a cent. Traditional banking rails, he said, do not support that degree of micropayment precision, and the fees and delays are too high. Crypto, by contrast, offers instant settlement, programmability, and the ability to compute in extremely small units, making it a better fit for machine-to-machine interactions.
That is where TRON fits into the picture in Sun’s telling. He described the TRON network as a high-speed, low-cost highway for global fund flows. Sun said TRON currently processes tens of billions of dollars in transactions every day and reaches annual settlement volume in the tens of trillions of dollars, serving as a general-purpose clearing base for high-frequency, low-cost transactions for global users and, eventually, the machine world.
Asked whether that means financial systems that do not integrate blockchain will be unable to plug into a silicon-based economy, Sun said yes. As the agentic economy grows, he argued, more human economic activity will be delegated to agents. Any financial system that machines cannot natively read and settle will be pushed out, while programmable money built on blockchain, including stablecoins and smart contracts, will form the financial base layer of a hybrid silicon-carbon economy.
“Freedom first, wealth second” and the idea of Small Ego
The final stretch of the conversation returned to financial freedom. Alan referred back to the starting point of the podcast “The Road to the Revolution of Financial Freedom” a decade ago and asked Sun how his understanding has changed. Sun summed up his current view in six Chinese characters that mean: freedom first, wealth second.
He said many people hold the mistaken idea that freedom comes only after enough money has been made. In his view, the sequence runs the other way. Freedom begins as a mental unbinding and a high degree of mobility: not being trapped by fixed assets, not being captured by public opinion, and being willing to move across sectors and geographies in search of arbitrage. Wealth, he said, is a byproduct that follows.
To maintain long-term freedom and wealth, Sun referenced the “Small Ego” idea associated in the conversation with ByteDance founder Zhang Yiming. He gave it two layers of meaning.
The first is breaking through arrogance and the resistance created by vested interests. Once people succeed in a given version of the world, he said, they often begin defending the old system. That can show up in established figures dismissing fitness or mocking crypto. The more successful a person has been, the larger the ego can become, and the easier it is to turn into a defender of an outdated version.
The second is patching fast. When someone discovers a flaw in their own thinking, Sun said, they should absorb the new theory and update immediately. He used himself as an example, saying he had previously held mistaken views on muscle health and then changed course after accepting a different framework.
On how Small Ego helps founders and investors avoid being eliminated by a changing cycle, Sun compared real life to “Earth Online,” a game with constant version updates. Real estate, he said, once looked like a dominant character but was later weakened by a version change, while AI has now become a core mechanism. No one can rely on one move forever. The deeper obstacle to adaptation, he argued, is often not a lack of capability but the ego built from past success. The answer, in his view, is to keep updating one’s own operating system like software, stay mobile, and remain willing to adjust or rebuild in response to external data and facts.
TechFlowPost noted that the article reflects the guest’s personal views and a compilation based on public-source material, and does not constitute investment, medical, or legal advice.
