AI Agents Are Coming and They Want Your Wallet

In Brief

The Hack Seasons panel discussed the transition from a theoretical phase to practical use cases of AI and blockchain, highlighting both potential friction points and opportunities.

AI and blockchain, two of the most disruptive technologies of our time, have long danced around each other in theory. But during the Hack Seasons panel “AI Gets Real: From Hype to Hardcore Use Cases”, it was clear: the honeymoon phase is over. Now comes the hard part, actually making it useful.

Moderated by Preetam from QuillAudits, the panel quickly broke away from buzzwords and drilled into the real friction points and opportunities in combining AI with decentralized systems. Each panelist offered their perspective, some optimistic, others skeptical, but all grounded in practical experience.

Why AI Doesn’t Need Blockchain, But Might Depend on It Anyway

Michael Heinrich, Co-founder and CEO of 0G Labs, made a bold yet nuanced claim: “Technically, AI doesn’t need blockchains. But if we want a safe AI, one we can audit, verify, and trust, we do.” At 0G, Michael and his team are building what they call the largest “AI Layer 1” blockchain, training models with over 100 billion parameters in a fully decentralized and verifiable manner.

Why does that matter? Because without transparency, AI becomes a black box: opaque, centralized, and potentially dangerous. Michael cited real-world examples where AI systems have gone rogue, from cloning themselves to issuing threats. With blockchain, he argued, we get provenance: traceable data, decentralized training, and transparent model outputs. If AI is going to become part of our lives, and it already is, verifiability isn’t optional, it’s existential.

Mary Tran, CEO of Orochi Network, shared a similar perspective but added a crucial layer: user indifference. “Most users don’t care if their AI outputs are verifiable, yet,” she said. “But we’re building the future where they’ll need to.” Orochi is working on verifiable data pipelines, enabling not just data but also proof of caring data. In a world where AI agents interact with other agents and smart contracts on your behalf, Mary believes it’s essential that those interactions are rooted in trustworthy, auditable data. “Otherwise,” she warned, “you’re giving power to something you can’t understand, and can’t stop.”

The Problem with AI Tokens and the Hype Cycle

Kevin Lee, CBO at Gate, brought humor and realism to the conversation. “The only thing I’ve done in AI is beat Super Mario World with a neural net,” he joked. But then he got serious. “Let’s be honest, most AI tokens are just wrappers around ChatGPT. They’re hype.”

As a major exchange, Gate lists these tokens because markets demand them, but Kevin didn’t hold back: “It’s unfortunate that crypto is the only industry trying to justify AI-token marriages that often don’t make sense. Just because you can create a token doesn’t mean it solves a real problem.”

However, he acknowledged that AI does have meaningful applications in Web3. From fighting fraud in KYC processes to optimizing trading strategies and market surveillance, AI is already baked into Gate’s operations. The future, he argued, isn’t about flashy AI tokens, it’s about infrastructure, security, and consistency. That’s where the next wave of real value will come from.

What’s Actually Being Built That Works

The panel wasn’t all theory. There were clear examples of builders doing the hard work of making AI and blockchain function together.

Michael from 0G Labs highlighted one standout project on their chain: Jane, a protocol developing a combined AMM and perpetual DEX that uses AI agents to rebalance LP positions, identify high-yield tokens, and even act as autonomous liquidity providers. “DeFi is the first real frontier,” Michael said. “Imagine an AI agent that can scan 200 chains, identify the best yield opportunities, and execute trades for a user, autonomously, trustlessly.”

Jong Park from BNB Chain agreed on the importance of real-world impact. He cited protocols like Alaya AI and Stadium Science that use AI to crowdsource and verify datasets, from sleep cycles to user responses. And while BNB Chain brands itself as a general-purpose Layer 1, it’s been actively promoting itself as an “AI-first” blockchain, fostering tools that let users retain ownership and monetization rights over their data.

That, Jong emphasized, is the key to adoption. “When users can own and earn from their data, everything changes.”

The Verdict? It’s Still Early, but It’s Finally Getting Real

There was agreement on one thing: we’re early. Really early. But also, for the first time, it feels like we’re moving past the superficial stage.

Mary offered a vision where humans no longer directly interact with smart contracts, but instead rely on AI agents to handle everything. Preetam reminded everyone that security must scale with automation, and that the careless use of LLMs to write smart contracts is already leading to vulnerabilities.

The take-home message? Hype can’t be the business model anymore, whether it’s verifiable AI agents, decentralized data economies, or autonomous liquidity providers, the real value is being built slowly, quietly, and it has nothing to do with the next token pump. At the Hack Seasons Hanoi, this panel proved that AI gets real when builders stop selling the dream and start shipping the infrastructure.