In this third episode of LazTalks, leaders from across the Web3 AI space came together to ask a pressing question: is Web3 truly ready for AI? Hosted by Liam, CT Lead at Meits, the session brought together Elena Sinelnikova Metis Co-Founder and LazAI Advisor, Dr. Zehua Wang, Chief Advisor to LazAI, Zhe from DePHY, Ducking from Duckchain, and Joshua from zkPass. The conversation explored infrastructure readiness, agent governance, data ownership, and what still needs to be built.

‍

Setting the Stage for AI in Web3

Host Liam opened by introducing the speakers and framing the discussion around the infrastructure and trust required for AI agents and tools to thrive in decentralized environments. Elena Sinelnikova shared her perspective on how Metis and LazAI view the future of AI in Web3. She stressed that AI agents will only be impactful if the underlying blockchain layers can handle their speed, privacy, and compute demands while keeping ownership and rewards with the users. This aligns directly with LazAI’s mission to deliver a full-stack decentralized AI network with DATs (Data Anchoring Tokens), programmable AI agents, and monetization flows.

Building AI-Specific Web3 Infrastructure

Dr. Zehua Wang made the case for Layer 2 networks purpose-built for AI. While general L2s handle transaction scaling, AI-focused ones need high throughput and specialized execution environments. Zehua highlighted sandboxing AI models before deployment to prevent harmful outputs and ensure safe operations. This approach, he said, would help developers test behaviors and address potential security gaps before agents interact with live environments. These ideas are embedded in LazAI’s technical roadmap, where secure compute environments and proof-based verification are core components.

Governance and Reputation for AI Agents

Zhe from DePHY discussed the structural needs for AI agents to participate in Web3 governance. Agents will require voting rights, smart contract-based decision-making, and onchain reputation systems. He likened it to creating credit ratings for agents, where data from both onchain and offchain sources can establish credibility and trust. DePHY’s work on decentralized physical infrastructure could integrate with such governance frameworks to allow real-world devices and agents to build onchain reputations.

User Trust and Privacy as Core Principles

Ducking, Chief Duck Officer at Duckchain, reinforced the importance of building trust frameworks for AI agents. He pointed out that agent reputation should be tied to performance history and verifiable records, which could be critical for marketplaces, e-commerce, and user-facing services. Duckchain, a Telegram-based AI bot platform, is well-placed to apply these trust and reputation mechanisms directly within chat-based interactions where users already spend their time.

Privacy-Proofed Identity and Data Ownership

Joshua from zkPass brought the conversation to the role of privacy-preserving technology. He noted that any AI system in Web3 must give users control over their private data while still allowing agents to function effectively. He called on developers to focus on privacy-first approaches that also allow for transparent proof of agent actions. zkPass’s privacy verification protocols are designed to fit seamlessly into AI systems that require both proof and discretion in handling sensitive data.

What Comes Next

The panel left listeners with a clear sense of urgency. To make AI agents in Web3 viable:

• Crypto networks must adapt to AI’s speed, privacy, and compute needs
• Governance, reputation, and data ownership systems need to be designed for autonomous actors

The discussion made it clear that Web3 is on the verge of supporting AI agents at scale, but only if the ecosystem moves fast. As Elena put it, the work ahead is both technical and community-driven. Join the conversation and get involved by joining the LazAI Telegram and builders feel free to learn more in the Developer Discord.