Hook
Volume is the only truth the market respects, and Tencent just dropped a bombshell with zero trading volume attached. On a quiet Tuesday, the Chinese tech giant announced the full rollout of Miora, an “AI creative agent” boasting memory, contextual understanding, and multi-agent collaboration. The news rippled through WeChat channels and crypto Telegram groups alike, but here’s the cold hard fact: there’s no on-chain proof, no benchmark, no open-source code. The market is already pricing in hype—but I’ve seen this playbook before. When the faucet runs dry, the dryers crack. And Miora, for all its promise, is a centralised black box dressed in the trendy clothes of “AI agent.”
Let me be clear: I’m not dismissing AI’s role in creative industries. I’m a financial engineer who spent 28 years watching protocols rise and fall. I know a controlled experiment from a production-ready product. Miora is the former—and the blockchain ecosystem should treat it as a signal, not a solution.
Context: Why Now?
Tencent’s Miora lands at a peculiar crossroads. The broader crypto market is in a bull run, with NFT floor prices climbing and AI token narratives exploding. Projects like Fetch.ai and Bittensor have shown that decentralised AI has real traction. Yet here comes Tencent, a $500B walled-garden giant, offering an “AI creative agent” that lives entirely on its own servers. No token, no DAO, no on-chain governance. Just a promise of better ad banners and viral video scripts.
The timing is no accident. In Q1 2026, Chinese regulators tightened rules on generative AI output, mandating content labels and safety audits. Tencent, with its compliance muscle, can afford to launch now while smaller AI startups scramble. But for the blockchain community, this raises a fundamental question: do we want our creative tools to be centralised under a single corporation’s censorship policies? Or should creative AI be trustless, transparent, and owned by the creators themselves?
Core: The Technical Skeleton — and Its Gaping Holes
Let’s dissect Miora’s claimed capabilities based on the limited public information. Tencent says Miora is a “multi-agent collaborative creative agent” with memory, understanding, and planning. That sounds like an LLM-powered agent system—likely built on Tencent’s Hunyuan foundation model. But here’s what’s missing from the announcement:
- Architecture: Is it a planner-executor loop? A graph-based orchestration (like LangGraph)? No details. Without knowing the underlying coordination mechanism, we can’t evaluate its reliability or attack surface. In blockchain terms, it’s like announcing a Layer 2 without revealing the fraud proof or validity proof.
- Memory: Short-term working memory or persistent long-term memory? If it’s the former, each session starts from scratch—making it useless for brand-consistent creative workflows. If it’s the latter, where is that data stored? On Tencent’s servers, of course. No encryption model, no user-controlled keys. For any creator dealing with proprietary intellectual property, this is a non-starter.
- Multi-Agent Collaboration: The phrase “multi-agent” is thrown around loosely. Does Miora employ specialised sub-agents (e.g., one for image generation, one for text, one for video) that communicate via a shared context window? Or is it a single model fine-tuned to simulate multiple personas? The difference matters for trustlessness. A single-model approach is easier to audit; a swarm of black-box agents is a security nightmare.
- Modality Coverage: “Creative” could mean anything from text ads to 4K video. Which modalities are supported? Is there any on-chain verification of output authenticity? In a world where deepfakes and synthetic media are proliferating, Miora offers no transparency about provenance. Compare this to blockchain-based creative AI like Render Network’s AI rendering or Story Protocol’s IP registration—where every output can be timestamped and traced.
Based on my experience auditing AI token projects during the 2024-2025 cycle, I’ve seen this pattern before: a big tech company announces an “AI agent” with a flashy demo but zero technical specs. The market goes wild. Then the substance fails to materialise. Miora is currently at the “demo” stage. The real test will come when independent researchers get hands-on access—if they ever do.
Quantitative Reality Check
Let me run some numbers. Tencent reported 2025 revenue of approximately ¥600 billion. Even if Miora captures 1% of their ad business (which is around ¥200 billion), that’s a ¥2 billion incremental uplift—a rounding error. For comparison, a mid-tier NFT collection like CryptoPunks can do ¥2 billion in secondary volume in a single month during a bull run. The market is overestimating Miora’s revenue impact.
On the cost side, multi-agent creative generation is computationally expensive. A single task (e.g., “generate a 30-second video ad for a luxury watch”) could require multiple model invocations: text prompt, image generation, video generation, voiceover, and iterative refinement. At current GPU pricing on Tencent Cloud, that could easily cost ¥5-10 per task. If Tencent faces mass adoption among WeChat’s 1.2 billion users—say 10 million tasks per day—that’s ¥50-100 million daily inference cost. Even for Tencent, that’s significant. Without aggressive optimisation (quantisation, speculative decoding, dedicated ASICs), Miora could be a cash sink. The volume is the only truth the market respects, but here the volume of compute might exceed the revenue.
Contrarian: The Unreported Angle
While everyone focuses on Miora’s potential to disrupt creative industries, I see a different story: Miora is a honeypot for regulatory compliance data. Every prompt, every generated image, every user feedback—it all flows to Tencent’s servers. For a company that already sits on the world’s most granular social graph (WeChat), Miora is a massive new data pipeline. With China’s new AI content regulations requiring real-time monitoring, Tencent can train censorship models on every creative output. From a blockchain perspective, this is the antithesis of transparency.
Look at the parallels with controversial projects like Worldcoin: centralised identity and data collection wrapped in a utopian narrative. Miora is no different. It lures creators with convenience, then locks their creative data into a proprietary system. If you’re a blockchain builder, you should be asking: how can we build a decentralised alternative that Miora’s users can migrate to?
Another blind spot: the single point of failure risk. What happens when Tencent’s content moderation model incorrectly flags your creative campaign as violating policy? No on-chain appeal. No community governance. Your entire ad budget wasted. Compare this to a DAO-based creative agent where governance tokens let creators vote on moderation rules. Miora is a command economy; blockchain is a market economy.
Second-Order Effects
In the next 12–24 months, I predict: 1. Short-term (0-6 months): Miora will launch as a free tool inside Tencent’s ad platform, attracting small merchants. Expect flashy marketing case studies showing 3x productivity gains. But no independent audit. 2. Medium-term (6-18 months): A backlash from professional creators will emerge. Concerns about copyright, data ownership, and lack of customisation will drive a wave of interest in decentralised AI creative agents. Projects like MossAI (a hypothetical blockchain AI agency) will gain traction. 3. Long-term (18+ months): Regulators in Europe and the US will scrutinise Miora’s data practices, potentially blocking its use for cross-border advertising. This will open a window for decentralised alternatives that comply with privacy regulations by design.
Takeaway
Chasing ghosts in the digital art auction house has been the crypto industry’s favourite pastime. Miora is another ghost—a tantalising vision of automated creativity that lacks the transparency and trustlessness our industry fights for. Don’t get me wrong; AI agents are coming to blockchain, but they will be built on open protocols, not corporate APIs. The question is: will we let Tencent define the narrative, or will we build the on-chain rails for creative AI? Volume is the only truth the market respects, and so far, Miora’s volume is zero. Watch the data, ignore the press releases. When the faucet runs dry, the dryers crack.