Now

当前状态面板What I am working on now

看我最近真正投入的方向。The projects and questions receiving my real attention.

笔记、工具、项目与练习沿着月相和根系汇成成长树的水彩铅笔画 — 现在进行时 · 每一个节点都仍在向上生长In progress — every node is still growing upward

In Progress

Current Focus

Pawin 企业 AI 创业探索：作为联合创始人，把 AI 工具、真实需求、企业流程和产品原型结合起来，不只停留在 Demo。/ Co-building enterprise AI products as a Pawin co-founder.Pawin enterprise AI: co-building practical products that connect AI tools, real needs, enterprise workflows, and working prototypes.
结构化认知链路：连接 video-AI-eye 与小叶认知系统，把碎片内容转成可核验、可确认、可检索、可复用的认知数据。Structured cognition pipeline: connecting AI Eye and the Xiaoye Cognitive System so fragmented content becomes verifiable, confirmable, searchable, and reusable.
申研 / 深造准备：整理项目、研究兴趣和未来方向，公开部分只保留方向和阶段。Graduate-study preparation: organizing projects, research interests, and future directions while keeping the public version appropriately scoped.

Shipping prototypes

Building

AI 眼睛：从用户主动打开的视频与文章中保留原始内容，按章节归纳并生成可编辑、可追溯的文字知识卡。AI Eye preserves source content from videos and articles selected by the user, then creates chapters and editable, traceable text cards.
小叶认知系统：把来源先编译为候选认知，经过人工确认和证据门禁后，再进入本地 Wiki、检索与受控行动。The Xiaoye Cognitive System compiles sources into candidate cognition, then applies human confirmation and evidence gates before local wiki, retrieval, or action use.
Codex Harness / OS Brain：研究 AI Coding 里的记忆、上下文压缩、Token 成本和个性化使用体验。Codex Harness / OS Brain explores memory, context compression, token cost, and personalized AI coding workflows.
高中政治 AI 备课系统：基于真实老师需求，结合飞鸽系统 Agent、本地数据库和云盘数据库持续迭代。The high-school politics lesson-preparation system is evolving from real teacher needs, agent workflows, and local and cloud knowledge sources.

Active notes

Researching

不同来源的碎片上下文如何统一为来源、原子认知、复合框架三层结构。How fragmented context from different sources can share a three-layer structure: source, atomic cognition, and compound framework.
如何让认知沉淀保留证据、适用边界和证伪条件，而不是只留下 AI 摘要。How cognition can retain evidence, applicability boundaries, and falsification conditions instead of becoming an unsupported AI summary.
AI Coding 工具的长期记忆和上下文压缩机制。Long-term memory and context compression for AI coding tools.
AI 教育产品如何围绕教材、时政案例、课堂活动和老师资料库形成工作流。How AI education products can connect textbooks, current-affairs cases, classroom activities, and teacher-owned materials into one workflow.

Organizing

Community

哈尔滨黑客松高校负责人：推进哈工大、哈工程、东林、东农四校联合高校黑客松。Coordinating a four-university Harbin hackathon across HIT, HEU, NEFU, and NEAU.
参与筹办面向社会群体的冰城哈客松，预计九月举办。Helping organize the Ice City Hackathon for the wider community, currently planned for September.

Preparing

Applying / Studying

把项目经历整理成可讨论的研究兴趣和成长证据。Turning project experience into discussable research interests and evidence of growth.
持续记录从设计学背景转向 AI 产品与工具链实践的路径。Continuing to document the path from design studies into AI products and toolchain practice.

最近更新Recent Updates

持续实践中的简短更新。Short updates from an ongoing practice.

新增 AI 眼睛与小叶认知系统：从视频和文章的碎片输入，走向经过确认的结构化认知与框架资产。Added AI Eye and the Xiaoye Cognitive System, connecting fragmented video and article inputs to confirmed structured cognition and framework assets.

补强 Lay 的成长手册：项目、笔记、工具和路径统一到浅色展览式视觉，并强化中英双语和创业探索人设。Strengthened Lay's Growth Manual with one exhibition-style system for projects, notes, tools, journey, bilingual reading, and venture practice.

把 Codex Harness / OS Brain、Vibe Copilot、高中政治 AI 备课系统和 Pawin 企业 AI 创业探索列为核心实践证据。Established Codex Harness / OS Brain, Vibe Copilot, the politics lesson-preparation system, and Pawin enterprise AI as core practice evidence.