Tencent's Miora: A Centralized AI Agent in a Decentralized World

Tencent launched Miora this week. An 'AI creative agent' with memory, need comprehension, and multi-agent collaboration. The press release is sparse. That silence is data. Every line of code writes a history of power. This one writes a history of centralized control, wrapped in the language of empowerment. We didn't ask for another walled-garden AI tool. But here we are. Miora is positioned as a creative assistant for advertising, embedded in Tencent's ecosystem—WeChat, QQ, Tencent Ads. It uses the Hunyuan model. It claims multi-agent orchestration. In the blockchain world, multi-agent means composability, permissionless verification, open audit. Here, it means proprietary orchestration behind a corporate firewall. The contrast is not subtle. Let's deconstruct Miora using a forensic lens. I've audited fifteen ICO smart contracts. I know that transparency is the baseline for trust. Miora offers none. No whitepaper. No open-source code. No third-party audit. The technical architecture is a black box. Is it a planning-execution-reflection loop? A simple LLM with function calls? A swarm of specialized agents? We don't know. The marketing says 'multi-agent collaboration'—but without open standards, that phrase is meaningless. It could be a single model with prompt engineering. It could be a genuine modular system. The lack of disclosure is a red flag. When I structured Aave's governance framework, we published the quadratic voting mechanism for public review. That's how trust is built. Tencent prefers opacity. That's by design. Miora's memory claims are another blind spot. Is it short-term context window or persistent vector database? Can users access, export, or delete their memory? In the Web3 world, data sovereignty is non-negotiable. In Miora, your creative history becomes Tencent's proprietary asset, fed back into their ad optimization loops. Governance isn't a feature. It's the foundation. Miora has no governance. The commercialization path confirms the walled-garden strategy. Miora will first serve Tencent's ADQ advertising platform. It's a cost-saving tool for advertisers, not a user-owned creative agent. The pricing model is unknown, but it will likely be consumption-based, tied to ad spend. This is not a platform that empowers individual creators. It's a tool that reinforces the advertising duopoly. Compare this to decentralized alternatives: imagine an AI agent that you own, that you train on your data, that you can rent out or sell outputs via smart contracts. Miora is the opposite. From an ethical and safety perspective, Tencent will enforce content compliance through its centralized review system. That's necessary for regulatory compliance in China, but it also means censorship and lack of recourse. The agent will generate content that aligns with Tencent's values, not the user's. There's no user governance. No on-chain dispute resolution. No transparency about the training data. The risks of bias, infringement, and manipulation are managed by a single corporate entity. Truth emerges from transparency, not from silence. Miora is silent on these matters. Infrastructure-wise, Miora runs on Tencent's private cloud. It uses Hunyuan inference clusters. The compute is not decentralized. There is no verifiable execution. No proof that the agent is doing what it claims. In the AI-crypto convergence space, we talk about verifiable inference, zero-knowledge proofs for model outputs. Miora has none of that. It's a closed system optimized for Tencent's bottom line. Now, the contrarian angle. Some might argue that Miora's multi-agent collaboration is technically advanced, that it could set a new standard for creative AI. But I argue the opposite. Without open standards, without the ability for users to inspect, fork, or compose agents, the 'multi-agent' claim is just marketing. The real innovation would be to put this on a permissionless network, to allow users to own their creative data, to have governance over the model's behavior via DAO structures. Tencent didn't do that. They built a better cage, not a better ecosystem. In the current sideways market, investors are desperate for signals. Miora is not a signal. It's noise. But it reveals a broader trend: big tech is trying to absorb AI creativity into their platforms, just as they absorbed social media, search, and commerce. This is the opposite of the Web3 ethos. We need open, permissionless, verifiable AI agents that users control. Miora is a step backward. Based on my experience designing DeFi governance, I know that structure creates freedom. Miora's structure is designed to extract value, not to empower. The tokenization of creative assets? Not here. The ability to migrate your agent to another platform? Not here. The right to audit the model's decisions? Not here. So what does Miora mean for the convergence of AI and crypto? It means the battle lines are drawn. On one side, centralized AI platforms with closed agents, massive user bases, and regulatory capture. On the other, decentralized AI with open agents, user sovereignty, and verifiable trust. The winner will not be determined by technical prowess alone, but by philosophy. By whether we believe that creativity should be owned by corporations or by individuals. Every line of code writes a history of power. Miora writes a history of centralized control. The question is: will we accept that history, or will we write our own? We need more than a press release. We need transparency. We need open architectures. We need governance. Until then, treat Miora as what it is: a proprietary tool, not a revolution. The real revolution is still being built, by those who understand that governance is the ultimate user experience.