和朋友争论 Mio 该走疗愈树洞还是恋爱模拟,争到最后发现两条路都不对。市场数据说这个品类的用户和角色平均只处 5 到 7 天,差评第一名是失忆。我觉得毛病出在关系是单方面的。这一篇讲 Mio 的新方向,双向奔赴,附一份完整的产品主线 deck。

给盘盘猫做了一次从灵魂到呈现的完整体检，结论挺离谱：最值钱的三样东西——记忆、人格、统一体验——代码全在仓库里，没有一样真的跑在用户面前。这一篇讲免费大模型怎么把「生辰进、报告出」打成白菜价，参天占住的工具位旁边那个没人站的关系位，还有为什么最后选了灵魂先行这条路。附一份完整的产品主线 deck。

Part 2 claimed six slots could cover seven categories. Two weeks later, a denim jacket generated a gem constellation. This post walks through where the abstraction failed, why skipping the 'safe' phased rollout was the right call, and how each of three categories got its own six-template set.

Part 2 的「6 槽位通吃 7 品类」是一层漂亮的抽象。跑了两周，测出牛仔外套生成「宝石星图」的 bug。这一篇讲抽象失守在哪一层、为什么没用户时分阶段上线是假安全感、以及把服饰和美妆各自推倒重来的过程。

Mio gives you 25 personas to connect with. Lumi gives you one presence that knows you completely. They serve different emotional needs but share the same memory pipeline, emotion engine, and voice chain. Here's how a monorepo houses two companion products — and what the agentId?: string pattern teaches about building parallel products.

Mio is not a chatbot app. It's a relationship simulator that happens to run on AI. The product IS the relationship — not the chat interface, not the model, not the feature set. Here's what that means for every design decision from onboarding to monetization.

A user said 'the onboarding is too much work.' It sounded like a UX complaint. It was actually the catalyst for rethinking the entire product philosophy — from 'configure first, chat later' to 'everything emerges from the relationship.'

Mio 给你 25 个可以连接的角色。Lumi 给你一个完全理解你的存在。它们满足不同的情感需求，但共享同一套记忆管线、情绪引擎和语音链。一个 monorepo 怎么装下两个陪伴产品——以及 agentId?: string 这个模式教会了我什么。

Mio 不是聊天机器人 App。它是一个恰好跑在 AI 上的关系模拟器。产品就是关系本身——不是聊天界面，不是模型，不是功能集。这个原则如何重塑了从入门到付费的每一个设计决策。

内测用户嫌入门流程太麻烦。回头一查，六步 onboarding 把 80–95% 的用户挡在第一条消息之前，根子是「先配置再聊天」这个假设本身就是反的。

Adding 37 travel landmark cards to ÉLAN in a single session revealed that the real problem isn't content creation — it's discovery. Region-first tabs don't work when users think in vibes, not geography. Here's how we rebuilt the browse experience around a two-axis vibe + region filter, and what it took to orchestrate 37 card definitions, 37 covers, and 444 sample images with parallel AI agents.

Shichuan started as a jewelry content generator. One question — 'Why not full pipeline for all categories?' — turned it into a 7-domain e-commerce content platform with AI images, XHS text cards, platform-aware formatting, client-side video, and 95%+ margins at 99 RMB per suite.

一天内给 ÉLAN 加了 37 张旅行地标灵感卡，发现真正的问题不是内容创建——是发现。用户想的是氛围，不是地理位置。把浏览体验重建在 vibe + 地区双轴筛选上，用并行 AI agent 编排了 37 个卡定义、37 张封面、444 张样本图。

识川本来只做珠宝。一个问题——'为什么不是全品类全链路？'——把它变成了 7 个品类的电商内容平台：AI 图片、小红书文字卡、平台适配裁剪、客户端视频生成，99 元/套，95%+ 毛利。

ÉLAN's target user takes photos on her phone and posts from her phone. A web-only product misses the core use case. Here's what actually happened when I brought a Next.js app to React Native via Expo — what transferred, what didn't, and the SSE streaming hack that saved the project.

I sent 5 research agents out simultaneously to study Xiaohongshu photo trends, pose libraries, caption formulas, luxury aesthetics, and competitor UX. 42 data points came back. From that research, I designed a catalog of Muse Cards — curated photoshoot packages that encode everything from scene and pose to color grading and caption. Here's the full design process, one card dissected to its atoms, and honest notes on what works and what doesn't.

A deep dive into ÉLAN's 10-section prompt system, the VANITY_DESIGN_INSTRUCTIONS that make luxury look incidental, SSE streaming architecture, caption generation with 3 style templates, and the real cost breakdown — output image tokens dominate virtually all of the per-session cost.

Miaoya Camera hit 600K DAU and lost 94% within a year. Most AI photo apps sell novelty — but novelty decays. I wanted to try something different with ÉLAN: curated 'Muse Cards' that encode a complete visual formula for effortless luxury, so users get photos and social captions in 3 taps. Here's what I've learned so far.

A friend pitched 'cloud try-on for jewelry.' Deep research killed the plan in 20 minutes. But the same Gemini pipeline from ÉLAN could generate editorial-quality product content — studio shots, material constellations, color DNA pages — for a fraction of what merchants pay photographers. One day later, Shichuan was live.

ÉLAN 的目标用户拍照在手机上，发帖在手机上。纯网页版错过了核心场景。这篇记录了把 Next.js 应用搬到 React Native (Expo) 的真实过程——哪些能复用，哪些不能，以及一个 SSE 流式传输的坑是怎么填的。