Luke Bowman, known as AnewbiZ and COO at Secret Network, moderated a panel at Futurist in Miami featuring Andrew Sato, founder and CEO of OurGen7 AI; Shaw Walters, founder of Eliza Labs; Janet Adams, COO of SingularityNET; and Patrick Delaney, CEO of Ampli. Bowman opened by framing the session as a convergence story. “We’re standing at the nexus of decentralized computation and autonomous intelligence,” he said, pointing to “the immediate innovation, the inevitable risk and the massive opportunities that the convergence creates.”
He anchored that convergence in a critique of where power sits today. “The goal of this frontier is to tackle centralization of data, computing and control by merging the power of AI with the control of web3,” Bowman said, before positioning Secret Network’s focus as a practical enabler. “At Secret network, we focus on confidential computing and the ability to secure sensitive data while still using it for things like inference and training models.”
Andrew Sato brought the conversation down to the physical realities of scaling AI. His emphasis was that the bottlenecks are increasingly about compute supply and power demand, and that decentralised infrastructure is part of how the industry will respond. Bowman acknowledged the constraint directly. “Yeah, the GPU market is a big one,” he said.
Shaw Walters took a different angle, arguing that decentralised AI will not scale through technology alone. For him, it rises or falls on coordination: aligning incentives, identity, performance and trust so that agents can compete and improve in visible ways. He described Babylon as a structured environment for that work. “So right now, we’re working on a game called Babylon,” he said, tying it to emerging standards work and the need for shared registries.
Walters broadened the point beyond agents and into general infrastructure. “I think bigger than that, even people who aren’t in the agents, it should just have the registry of all the applications on your chain, to have a reputation system,” he said, outlining the kind of layer that lets users and other agents evaluate reliability at speed.
Patrick Delaney kept the focus on trust and security design. His argument was that systems should separate what an agent recommends from what a user authorises and what a system executes. “Separating agents, intent and suggestion from the user specified policies makes a ton of sense,” he said.
He also described the behavioural gap decentralised systems must bridge. “As long as the servers are centralized, we must trust the company,” Delaney said, capturing the default comfort users take from a known operator. Bowman linked that back to Secret Network’s positioning. “We actually work with confidential computing, protection and security of data and right there with you,” he said.
Janet Adams made the case that current infrastructure is not built for the AI workload now arriving. “Today’s AI based on large neural networks is extremely resource hungry,” she said, and argued that the solution cannot be bolting AI onto blockchains that were designed for simpler execution. “We also need blockchain that’s actually built for AI,” she said.
Adams also pushed hard on standards and interoperability. “We all need to collaborate with each other much more,” she said, calling for shared “standards for agents” so that “my agents and your agents can actually work together.”
In conclusion, Bowman’s message was that decentralised AI is moving from experiment to infrastructure. If it cannot handle sensitive data, integrate across projects, and remove friction for large organisations, it will stay niche. If it can, it stops being “crypto AI” and becomes part of how business is done.
