Beyond his work as an author and thought leader, Lian serves as Chief Digital Advisor to the Mongolian Productivity Organisation and is a partner and fund manager overseeing blockchain investments for Passion Venture Capital Pte. Ltd. An early blockchain adopter, investor, and entrepreneur, he has advised governments, public companies, and organizations across Asia on digital assets, emerging technologies, and innovation strategy. He previously served as Chairman of BigONE Exchange and as an Advisory Board Member of Hyundai DAC, the blockchain arm of Hyundai Motor Group.

Your career has taken you from early Bitcoin investing and blockchain advocacy to advising governments, regulators, and enterprises across Asia. Looking back over the past 15 years, what experiences most influenced the ideas that ultimately became Web4: The Age of Autonomous Intelligence? 

When I first encountered Bitcoin back in 2012, people dismissed it as a passing curiosity. But sitting in rooms with early tech pioneers, I saw the beautiful genesis of decentralized trust. Later, bridging the gap between that wild frontier and institutional stability, whether advising the Asian Productivity Organization, think tanks, or various ministries across Asia, I was slapped with a cold dose of reality.

I watched regulators panic over crypto’s lawlessness, while tech purists ignored the blindingly obvious fact that standalone smart contracts were functionally rigid and lacked true intelligence. The defining trigger for Web4 was observing this massive fragmentation firsthand. I chaired DAOs where a single wealthy whale’s vote effortlessly overrode thousands of passionate community members. I watched brilliant, everyday people look at Web3 interfaces like they were trying to decipher alien hieroglyphics. It became glaringly obvious to me that you can’t achieve true digital sovereignty if the system’s cognitive burden remains entirely on a human user who is already exhausted. That friction is what planted the seed for Web4.

In the book, you argue that Web1 gave us information, Web2 gave us connectivity, and Web3 gave us ownership, yet all three failed to fully solve the problem of digital sovereignty. What is the fundamental flaw in today’s internet architecture that Web4 is designed to address?

The absolute fundamental flaw in today’s architecture is what I call the “centralization paradox”. Web2 was essentially history’s greatest bait-and-switch; we gladly traded our intimate behavioral data for the convenience of “free” tools, turning ourselves into the product for corporate landlords. Web3 arrived waving the flag of digital liberation, screaming that we could finally own our digital assets.

But let’s look past the slick marketing : Web3 fundamentally failed because its underlying infrastructure fell right back into the hands of specialized elite, VC firms, and concentrated validator networks. If you need a centralized exchange or an industrial cloud provider just to interact with a “decentralized” network, you aren’t sovereign; you’re just renting space from a different landlord. Web3 gave us the immutable ledger for ownership but left the punishing cognitive load completely on the user. Web4 fixes this structural decay by introducing a cognitive layer, allowing automated systems to process this heavy structural friction on our behalf while keeping everything strictly verified on-chain.

One of the central themes of the book is that Web4 emerges from the convergence of artificial intelligence and blockchain. Why do you believe these technologies are more powerful together than separately, and what new capabilities become possible when AI serves as the cognitive layer and blockchain serves as the trust layer?

For too long, the tech world treated AI and blockchain like separate, competing silos. AI developers mocked blockchain for being slow and rigid, while blockchain purists looked at AI as an untrustworthy, manipulative black box. They were both completely right, and they were both missing the grand architectural picture.

When you pair them, they become the brain and the spine of a completely new internet paradigm. AI provides the cognitive agency, the adaptable brain that can read natural language, reason through messy data, and make continuous decisions. Blockchain provides the unyielding spine of trust, the immutable audit trail that prevents that brain from going rogue or serving corporate profits. Together, they form a symbiotic feedback loop. You get magnificent new capabilities like fully autonomous AI economic actors that can manage finance or execute cross-chain agreements without a human operator, yet can cryptographically prove they acted perfectly within the user’s explicit parameters. It’s intelligence, completely verified.

You write that users should be able to express intent in natural language while autonomous AI agents execute tasks on their behalf. What technological breakthroughs still need to occur before this vision becomes practical for everyday consumers and businesses?

Right now, telling a chatbot to “optimize my financial portfolio” results in a nice essay, not execution. To bridge natural human intent with on-chain reality, we require a massive technological shift from reactive, prompt-based chatbots to true, persistent autonomous agent layer architectures.

First, we need to completely solve the determinism bottleneck. Blockchains require absolute, mathematical certainty across every node, whereas AI model inference operates heavily on probabilistic approximations and fluid numbers. Reconciling those two alien computational logic systems requires deep cryptographic breakthroughs. Furthermore, generating Zero-Knowledge Machine Learning (ZKML) proofs, which allow an agent to prove its decision-making logic was untampered without exposing sensitive parameters, currently takes far too long. We desperately need hardware acceleration and circuit optimization to slash proof-generation latency down to milliseconds. Finally, we need universal cross-chain intent standards, such as ERC-7683, so specialized agents can seamlessly coordinate transactions across a messy, fragmented multi-chain universe without getting trapped in siloed networks.

The book introduces the concept of AI economic actors that can operate independently, make decisions, allocate capital, and even run businesses. How realistic is that future, and what are the biggest governance and accountability challenges that society will need to solve along the way?

This isn’t sci-fi speculation; it is an economic inevitability. In the book, I forecast that by 2035, fully autonomous AI will command roughly 50% of all digital platform decision-making, an explosive jump from the meager 15% we see today. We will see autonomous businesses operating entirely on smart contracts, dynamically adjusting tokenomics, hiring human contractors, and optimizing yields without a human CEO in sight.

The real headache isn’t whether the tech works, but how we govern it. If an autonomous entity discovers a predatory, algorithmic strategy that maximizes protocol profit while completely draining a local community’s liquidity, who gets sued? Current corporate law completely breaks down when there is no human operator behind a legally binding transaction. We will need to design neutral, multi-agent validation frameworks and strict, hardware-enforced threshold safeguards to ensure machine intelligence serves human policy rather than cold, unaligned optimization functions.

Privacy is a recurring theme throughout Web4. Technologies such as Zero-Knowledge Machine Learning, Federated Learning, and Homomorphic Encryption play a prominent role in your framework. Which of these innovations do you believe will have the greatest impact on making AI both useful and privacy-preserving?

While all three are essential pillars of the framework, Zero-Knowledge Machine Learning (ZKML) is the absolute crown jewel of the Web4 trust architecture. Federated Learning is great for decentralized training without moving raw data, and Homomorphic Encryption offers bulletproof math but carries a brutal computational penalty that slows things to a absolute crawl.

ZKML solves the foundational crisis of the autonomous web: the black-box problem. It elegantly strips away the toxic binary choice of either blindly trusting a corporate AI provider or downloading an entire proprietary model. With ZKML, an agent can analyze highly sensitive personal inputs, like your private medical data or institutional treasury records and cryptographically prove that the resulting output was generated by the exact, unmanipulated certified model, all without exposing the model’s secret weights or violating the user’s data sovereignty. It transforms trust from a blind leap of faith into verifiable, mathematical proof.

You mention that this book took three years to complete and went through 23 versions before publication. What were the biggest changes in your thinking during that process, and how did the rapid evolution of AI influence the final version of the book?

Writing this book was a brutal exercise in chasing a supersonic target. When I first started mapping out Web4 concepts back in 2021, the text was primarily a sharp, technical critique focused heavily on solving Web3’s structural centralization and painful UI bottlenecks. It was much simpler back then.

But as generative AI and large language models aggressively went mainstream, my entire thesis flipped. I watched the technology landscape move rapidly from static, reactive chatbots to complex, goal-seeking agentic behaviors. I found myself constantly ripping out chapters because the real-world tech was moving faster than the ink could dry. I ended up cutting over 140 pages just to keep the architectural blueprint practical and human-centric. The rapid shift made me realize that Web4 isn’t just a simple technical patch for crypto wallets ; it’s an urgent global race to build a decentralized substrate for intelligence before corporate monopolies lock us into an inescapable era of ultimate surveillance capitalism.

You have spent considerable time working with policymakers and regulators. As autonomous AI systems become more capable and decentralized networks become more intelligent, how should governments balance innovation with oversight without repeating the mistakes that slowed adoption of earlier technologies?

Governments always default to the exact same mistake: trying to write industrial-era, rigid laws for technologies that mutate every single week. If you apply heavy-handed, legacy licensing compliance models to a decentralized network, the innovation simply packs its bags and migrates to a friendlier jurisdiction overnight.

The secret to modern regulation lies in changing the mindset from enforcing human paperwork to leveraging programmable, automated parameters. Regulators need to realize that Web4’s trust architecture is actually their greatest ally. Instead of demanding backdoors or centralized oversight, governments should collaborate on open technical standards like verifiable audit trails and ZK-proof validation. By encoding compliance guidelines directly into the protocol’s consensus pipeline, you allow real-time anomaly detection and transparent, neutral tracking without choking developers with red tape. The goal is to move smoothly from retroactive, heavy-handed legal punishment to continuous, cryptographic policy-making.

Digital securities have long been viewed as one of blockchain’s most promising use cases, yet adoption has been slower than many expected. How do you see digital securities evolving over the next decade, and what role could AI play in areas such as compliance, issuance, trading, and investor protection?

The adoption of digital securities hit a structural brick wall because traditional finance demands dynamic, real-time risk management and complex compliance logic that basic, static smart contracts simply cannot process.

Over the next decade, the integration of AI will breathe massive liquidity into this stagnant sector. AI agents will transform digital securities from static digital wrappers into living, intelligent assets. Imagine an issuance framework where AI oracles continuously parse unstructured global news feeds, supply chain shifts, and changing regulatory text, instantly updating the underlying security parameters or modifying cross-border compliance routing rules on-the-fly. For investor protection, specialized multi-agent swarms will monitor on-chain transaction patterns continuously, instantly detecting malicious spoofing or front-running attempts and triggering hardware-level freezes before retail capital is wiped out. AI will finally turn digital securities from a technical novelty into enterprise-ready institutional financial infrastructure.

If we fast-forward to 2035 and your vision of Web4 becomes reality, what will surprise people most about how they interact with the internet, and what aspects of today’s digital experience do you think will seem completely outdated?

By 2035, the concept of manually “browsing” the internet or clicking through rigid apps will feel as painfully ancient as using a dial-up modem or a fax machine. People will be completely shocked by the total disappearance of the digital middleman. You won’t navigate corporate websites to buy flight tickets, rent a car, or manage an investment portfolio ; you will simply speak your intent to your sovereign AI agent in plain, everyday language.

Our current digital nightmare, where we are constantly bombarded by toxic clickbait, trapped in addictive algorithmic dopamine loops, and forced to remember dozens of complex passwords and seed phrases, will seem completely medieval. Today, we are the extracted fuel for corporate machines ; in 2035, the web will function as a decentralized, deeply personalized extension of human intent. We will look back at the mid-2020s and laugh that we actually allowed a handful of centralized tech platforms to profile our private lives and sell our attention back to us.

Thank you for the great interview, readers should also consider reading Web4: The Age of Autonomous Intelligence.