More than 70 crypto and Web3 companies including big names like Yuga Labs, Magic Eden, and Arbitrum are teaming up to push back against Big Tech’s growing control over artificial intelligence.
The coalition is backing Thinkagents.ai, a new open-source protocol designed to let AI agents operate independently across decentralized networks, putting control back in users’ hands.
Unlike traditional AI systems, which lock data and functionality behind corporate walls, Thinkagents.ai is built on the Think Agent Standard (TAS), a framework that makes sure users own their AI agents, data, and interactions.
Mike Anderson, a core contributor to the project, compares it to the early days of social media, except this time, users wouldn’t surrender their data to platforms like Facebook or Twitter. Instead, AI agents would act as private extensions of the users that travel with them across apps without leaking or selling their information.
The protocol has already had a public test run. Last summer, a live experiment featuring AI agents playing Street Fighter 3 attracted around 30,000 viewers and participation from nine companies. It offered a glimpse into how interoperable, decentralized AI could work in practice.
At the heart of the system is Non-Fungible Intelligence (NFI), a digital identity layer that gives each AI agent a unique “Soul” (identity), “Mind” (decision-making), and “Body” (cross-platform functionality). The first consumer application, SOULS, lets users train and customize their own AI agents while maintaining full ownership.
Anderson warns that without decentralized alternatives, AI could repeat the mistakes of social media, where “your data exhaust is more valuable than your DNA.”
With early adoption from firms like Alchemy and Yuga Labs, Thinkagents.ai may be laying the groundwork for a more open, user-first AI ecosystem, one where the next generation of digital agents truly belongs to the people.
