On a quiet Tuesday, Tencent dropped a press release that barely rippled beyond the Chinese tech echo chamber.
Miora. An AI creative agent with memory, needs comprehension, and multi-agent collaboration.
A tool for generating ads, copy, and visuals inside the Tencent ecosystem.
Most analysts yawned. Another me-too product in a crowded field.
But I see something else. A signal. A deliberate move to lock the creative supply chain behind a central gatekeeper. And for anyone who has spent years reading the code that writes the culture — this is the exact pattern that preceded the ICO crackdown, the DeFi yield carpet pulls, and the NFT identity crisis.
The playbook is the same. First, a centralized entity builds a closed system that seems convenient. Then they control the rails, the data, and the value capture. Finally, they pull the rug on the open alternatives.
Navigating the storm to find the steady current means recognizing that Miora is not just a product. It is a paradigm. A centralized AI agent paradigm. And it is coming straight for the decentralized creative economy that Web3 has been nurturing.
Let me step back.
Context matters more than ever in a bear market where every protocol is bleeding LPs and every "proof of reserves" is a theater of liability. You need to know where the real battles are being fought.
The AI agent narrative has been simmering in crypto for two years. Projects like Autonolas, Fetch.ai, and numerai have built decentralized networks for autonomous agents. The promise: agents that operate on-chain, permissionlessly, owned by users, and earning tokens for their work.
But adoption is glacial. The user experience is fragmented. The compute is expensive. And the average crypto user still thinks an agent is just a chatbot.
Meanwhile, Big Tech has been quietly assembling the full stack. Google launched Gemini Agents. Microsoft Copilot is now a multi-agent system. And Tencent — with its 1.3 billion monthly active WeChat users and the largest advertisement platform in China — just dropped Miora.
Miora is not a research paper. It is a deployed, integrated product. It lives inside Tencent's advertising division, the very engine that generates ~30% of the company's $80 billion annual revenue.
That is not a side project. That is a strategic deployment of capital and compute to own the creative output of the world's second-largest economy.
Now, let's break down what Miora actually is — and why it matters for every builder in crypto.
Before I dig into the technical architecture, let me be blunt: Based on my experience auditing over 50 ICO whitepapers in 2017, I learned to spot the gap between marketing claims and reality. Miora's press release is thin. It mentions 'memory, needs comprehension, and multi-agent collaboration' but gives zero technical specs. Zero benchmarks. Zero open-source code.
That tells me two things. First, Tencent is treating Miora as a proprietary competitive moat. Second, the technology is likely a combination of existing models — the Tencent Hunyuan LLM, a diffusion model for images, and a video generation component — orchestrated by a simple planner-executor agent loop.
Multi-agent collaboration in this context almost certainly means a modular architecture: a planner agent that decomposes a user's request (e.g., 'create a Mother's Day ad for my tea shop'), delegates sub-tasks to specialized agents (copywriter, designer, video editor), and then merges the results. This is the same pattern used by MetaGPT, AutoGPT, and CrewAI.
But here is the critical difference: coordination is centralized. The planner agent is a black box controlled by Tencent. The memory is stored in Tencent's vector databases. The generated content is filtered through Tencent's content moderation pipeline. The entire system is rent-seeking by design.
Compare that to decentralized agent frameworks like Autonolas's OLAS stack, where agents are deployed on-chain, their code is open, and coordination happens via smart contracts. In a permissionless system, the value created flows to the agent's owner, not to a corporate intermediary.
Miora represents the opposite philosophy: an AI agent that works for the platform, not for you.
Now let's talk about commercialization.
Miora has no announced pricing. No API. No independent revenue. It is currently a free feature bundled with Tencent's advertising platform (ADQ) and likely serving as a data collection honeypot. Every ad brief, every image generation, every iteration trains the model and builds a dataset of creative preferences tied to specific advertisers.
This is the classic freemium trap. Once adoption reaches critical mass, Tencent will monetize through usage tiers, creative token limits, or priority queue fees. The advertisers become locked in because their creative assets and historical performance data live inside the platform.
From a crypto perspective, this is the equivalent of a centralized exchange that offers free trading for six months, then introduces withdrawal fees and KYC requirements. It is a known pattern. And it works.
But there is a deeper structural problem: Miora is designed to serve the advertising economy, not the creator economy. Its 'creativity' is constrained by what drives click-through rates and conversion. It will optimize for bland, safe, high-performing assets — not artistic expression, not novelty, not cultural commentary.
This is where the decentralized alternative has an opening. Tokenized creative agents can operate with different incentive models: agents that earn royalties for unique content, agents that collaborate with human artists on chain, agents that generate NFTs with provable provenance.
But that opening is narrowing. Tencent is investing heavily in multi-agent orchestration, and its compute resources dwarf any crypto-native project. Miora may not be technically superior to a decentralized agent, but it has distribution. And distribution beats technology in the short term.
Let me address the elephant in the room: the supposed 'competition' with ByteDance's Jichuang, Alibaba's Tongyi Wanxiang, and Baidu's Wenxin Yige.
From a Western perspective, this looks like a crowded market. But inside China, it is a winner-take-most ecosystem battle. The winner will be the one that integrates most deeply with the existing social and commerce infrastructure. Tencent has WeChat, QQ, and the largest advertising platform. That is a formidable moat.
For crypto projects, the competitive landscape is not Miora itself — it is the normalization of centralized AI agents that control creative inputs and outputs. If users become accustomed to a walled-garden agent that requires KYC, censors outputs, and takes a cut of every generated asset, they will not even think to demand a permissionless alternative.
I saw this happen with stablecoins. For years, the crypto native dismissed USDC and USDT as centralized tokens. But when the market crashed and on-chain liquidity dried up, those centralized stablecoins became the only game in town. The same pattern will repeat with AI agents.
The contrarian angle: Miora's launch is not bad news for decentralized agents. It is the catalyst that forces builders to prioritize usability and distribution over idealogy.
Let me pivot to a specific concern that keeps me up at night: the regulatory implications.
China's Generative AI regulations require all AI-generated content to be labeled, filtered, and traceable. Miora will comply by design. That means every image, every ad copy, every video generated will carry a watermark and be logged on Tencent's servers.
Now imagine a crypto project that wants to allow agents to generate and trade creative assets on-chain. If a regulatory body demands accountability, a centralized platform can provide it. A decentralized platform cannot — and that becomes a liability.
But here is the blind spot: The same regulation that ensnares decentralized agents also creates a market for privacy-preserving creative agents. Zero-knowledge proofs can verify that an agent generated a piece of content without revealing the input prompts or the user's identity. Projects like Aleph Zero and zkSync are already working on privacy layers for AI.
Miora cannot offer that. Its architecture is inherently transparent to Tencent. That is a product differentiation that crypto can own. But only if we build it before the regulatory window closes.
Now, let's talk about the numbers. In a bear market, every metric matters.
Miora is not a public company, so no valuation. But we can estimate the addressable market. China's digital advertising market is ~$140 billion annually. If Tencent captures 10% of creative production through Miora, that is $14 billion of revenue opportunity — albeit mostly as cost savings for advertisers, not direct revenue for Tencent.
Compare that to the entire market cap of all decentralized agent projects, which is under $5 billion. The asymmetry is staggering. But it also means the growth potential for decentralized agents is enormous if they can capture even 1% of the same market.
The key question: Can a decentralized agent platform offer a better value proposition than Miora for a small business owner in Shanghai?
Right now, no. Miora is free, integrated with their existing ad account, and requires no technical knowledge. A decentralized agent would require a wallet, a token payment, and trust in an opaque codebase.
But that can change. If a decentralized agent platform offers lower fees, actual ownership of the creative assets, and interoperability across platforms (e.g., use the same agent on WeChat and on a Web3 social network), the value proposition shifts.
This is the battleground for the next bull run: distribution versus sovereignty.
Let me share a hard-earned lesson from my years covering this industry. In 2022, after the FTX collapse, I wrote a post-mortem that got shared 100,000 times. The central thesis: centralization of infrastructure is a systemic risk. FTX was a centralized exchange that lied about reserves. Terra was a centralized algorithmic stablecoin that miscalculated risk.
Miora is not a fraud. But it is a centralization of the creative infrastructure. And every centralized infrastructure creates a single point of failure — for censorship, for data lock-in, for regulatory capture.
The crypto community needs to stop treating AI agents as a parallel track and start integrating them into the core value proposition of decentralization.
Projects like Bittensor are already doing this by creating decentralized networks for AI model training and inference. Agent platforms like Vana are focusing on user-owned data. But we need more. We need agents that can generate creative assets, mint them as NFTs on chain with ownership attribution, and trade them on decentralized marketplaces without a gatekeeper.
That is the vision. Miora is the anti-vision. And we need to treat it as such.
Now, allow me to address the inevitable skepticism: "But Emma, Tencent is just building for its own ecosystem. It doesn’t threaten crypto’s niche."
That’s exactly what people said about Facebook’s Libra in 2019. Look how that turned out. Libra never launched, but it triggered a regulatory backlash that set back decentralized stablecoins by years. Governments saw a global stablecoin and panicked. Miora is not a stablecoin, but it is a demonstration that AI agents can work at scale inside a closed system. Regulators will see it and start asking: "Why do we need permissionless agents when this works so well?"
We are not competing with Miora for users. We are competing for the narrative of what an AI agent should be.
Let me close with a concrete scenario.
It is 2027. A small business owner in Vietnam wants an agent to create ads for her e-commerce store. She has two options.
Option A: Sign up for Tencent’s Miora via WeChat (already installed). It costs nothing upfront. The agent generates high-quality images and videos in her local language. It offers detailed analytics and integration with her existing ad accounts. The catch: Tencent owns the training data, can censor her content, and takes a 10% platform fee on all generated output that leads to sales.
Option B: Download a decentralized agent app that connects to her Web3 wallet. The agent is open-source, audited, and runs on a decentralized compute network. She pays per inference with a token. She retains full ownership of the outputs and can sell them on an NFT marketplace. But the UI is clunky, she needs to understand gas fees, and the agent’s quality is slightly lower because the models are smaller.
Which option wins? Today, Option A. But it doesn’t have to. The gap can be closed with better UX, funding from a tokenized ecosystem, and most importantly — a narrative that resonates.
That narrative is what I write. And I will bet on it.
So here is my takeaway for the builders reading this.
Stop obsessing over which L2 has the lowest fees. Stop arguing about the next memecoin narrative. The real narrative is the battle for the creative agent layer. Miora is a wake-up call. It shows that centralized players are moving fast, spending billions, and integrating AI into their core profit centers.
The crypto response should not be to build an alternative, but to build a better one — one that is sovereign, composable, and aligned with user ownership.
We have the technology. We have the talent. What we lack is the sense of urgency.
Navigating the storm to find the steady current means seeing Miora not as a competitor, but as a benchmark. A benchmark for how good a centralized agent can be — and a reminder that the decentralized version must be even better.
I will be watching the metrics. And I expect you will be too.
Reading the code that writes the culture, I see the blueprint. The question is whether we will code our own future or rent it from Tencent.
