rewrite this title Insights From Hack Seasons Singapore:  Speakers Share How AI Can Accelerate Web3 Development

by
Alisa Davidson

Published: October 13, 2025 at 8:01 am Updated: October 13, 2025 at 8:01 am

by Ana

Edited and fact-checked:
October 13, 2025 at 8:01 am

To improve your local-language experience, sometimes we employ an auto-translation plugin. Please note auto-translation may not be accurate, so read original article for precise information.

At the start of October, the Hack Seasons Conference was held in Singapore, bringing together leading professionals to examine the latest trends in AI, Web3, blockchain, and cryptocurrency. One of the day’s most notable panels, titled “AI: The Logic and Intelligence Layer Web3 Needs?,” focused on how AI-driven analytics can enhance on-chain activity, creating faster, smarter, and more adaptive ecosystems. The discussion, moderated by Igor Lessio, CEO of AIFlow, featured prominent figures including Will Carter, CTO of Lamina1; Lane Retting, Head of Research at Near Foundation; Dillon Chen, Founder and CEO of Common; and Aaron Ramirez, AI Software Engineer at Flipside.

The discussion began with an acknowledgment that Web2 and Web3 differ in the speed of development, prompting the question of what currently prevents Web3 from achieving the same pace as Web2.

Panelists agreed that Web3 is moving slower than Web2, but emphasized that this is not necessarily a drawback—it is, in many ways, a feature. Unlike Web2, Web3 prioritizes security and finality. Building in Web3 requires careful implementation of guardrails to ensure that decentralization is preserved and that reliance on opaque systems, such as black-box large language models, does not introduce new risks. Web3 developers must solve unique challenges that Web2 does not even encounter.

The panel further highlighted that human systems in Web3 closely mirror technical systems. In a centralized Web2 application, developers can rely on a central database or oracle to achieve high throughput, fast transactions, and scale—but this comes at the cost of centralization. By contrast, using a blockchain in Web3 is slower and more expensive, but it provides immutability and other core decentralized features. The slower pace also extends to the social layer, where decentralized governance processes and community-driven decision-making contribute to better long-term outcomes, even if they slow development. AI agents, the panel noted, can help mitigate some of these inefficiencies by automating repetitive tasks or supporting governance, effectively accelerating certain aspects of decentralized processes.

Another point raised was that innovations in cryptocurrency space often remain invisible to end users. Mechanisms such as consensus protocols and social governance layers operate behind the scenes, much like developments in the open-source AI community, which can be months behind the latest cutting-edge research.

A recurring theme was security. Unlike Web2, where speed often takes precedence over security, Web3’s primary focus is security and reliability, which naturally slows product release cycles. Panelists suggested that Web3 teams can leverage AI tools strategically to streamline development and accelerate launches without compromising safety, thereby combining the benefits of decentralization with more efficient production.

Balancing Decentralization And AI: Privacy, Transparency, And Multi-Agent Systems In Web3

The conversation shifted to the topic of decentralization, a core principle of Web3, and how it intersects with AI, which often represents centralized intelligence. Panelists explored strategies for maintaining decentralization of intelligence while leveraging AI technologies.

Privacy in AI models remains a challenge, particularly when systems operate remotely. In practice, most developers prioritize performance over privacy for many applications. On-device AI presents one way to preserve user data for limited use cases, but devices like smartphones can only run models of restricted size, creating a trade-off between performance, capability, and privacy. The panel noted that the general public often prioritizes convenience over privacy, opting for platforms that may exploit user data, citing examples such as WhatsApp and TikTok. However, smaller, targeted tools—such as onboarding applications—can benefit from local AI frameworks that translate technical data into user-friendly, intent-based interactions. Despite these solutions, AI models remain largely opaque black boxes, and more transparent alternatives are needed.

Speakers emphasized the importance of open and verifiable model training, including cryptographic proofs that ensure AI pipelines and on-device inference processes are trustworthy. They admitted that two years ago they had been overly optimistic about the public’s concern for privacy and verifiability. Today, while mass awareness remains limited, there are market segments that prioritize privacy, representing a significant opportunity in a potentially trillion-dollar market.

Beyond privacy, panelists discussed current limitations of AI models, noting that even advanced systems like ChatGPT can produce inconsistent outputs for the same queries. User experience, multi-agent systems, and decision-making processes are key areas requiring improvement. On trust, the panel underscored the need for companies to provide transparency about the data AI agents use.

Looking ahead, the discussion suggested that reliance on a single AI provider is risky. Diversifying across multiple providers, including open-source models, is advisable. Several emerging companies are developing AI products that store no user data and utilize open-source models, creating the potential for increased competition and more decentralized, privacy-focused AI solutions over the next five years.

Concluding the discussion, panelists reflected on how user experience (UX) has evolved, noting a clear shift toward chat-based interfaces. They emphasized that the next major wave in consumer DeFi will be shaped by experiences similar to ChatGPT, where users can access information in fractions of a second. They further highlighted that this evolution will require rethinking operational approaches to fully embrace this new, streamlined UX, particularly as most interactions now occur on mobile devices.

The panel noted that users are increasingly conditioned to expect immediate responses, making both speed and accuracy critical. At the same time, they stressed the importance of user literacy, pointing out that people still need to understand how large language models (LLMs) work, including their limitations, and that outputs should always be verified.

Alisa, a dedicated journalist at the MPost, specializes in cryptocurrency, zero-knowledge proofs, investments, and the expansive realm of Web3. With a keen eye for emerging trends and technologies, she delivers comprehensive coverage to inform and engage readers in the ever-evolving landscape of digital finance.

More articles