Users lack an integrated product that seamlessly combines route planning, real-time environmental cues, and guided species discovery to facilitate meaningful, context-aware outdoor learning. This forces users to switch between disconnected apps, sacrificing the joy and immersion of the outdoor experience for fragmented information.

用户缺乏一款能将路线规划、实时环境提示与引导式物种探索无缝整合的产品，以支持有意义的、情境感知的户外学习体验。这迫使用户在各类孤立的应用之间来回切换，为获取碎片化信息而牺牲了户外体验的乐趣与沉浸感。

How might we seamlessly integrate nature education into outdoor activity without sacrificing the experience's joy and immersion?

我们如何在不牺牲体验乐趣与沉浸感的前提下，将自然教育无缝融入户外活动之中？

Outdoor explorers and enthusiasts seeking a more meaningful, educational, and interactive connection with the natural world.

寻求与自然世界建立更有意义、更具教育性与互动性联结的户外探索者与爱好者。

The design was driven by two successive Lean UX cycles, each structured by Think, Make, and Check steps. This approach prioritized efficiency and continuous user validation. The first cycle defined the problem and validated low-fidelity solutions, while the second refined the user experience and visual design.

设计以两轮连续的精益 UX 循环为驱动，每轮均按照思考、制作与测试三个步骤展开。该方法注重效率与持续的用户验证。第一轮明确问题并验证低保真方案，第二轮则深化用户体验与视觉设计。

To understand outdoor enthusiasts' needs regarding nature exploration and environmental data, I conducted user interviews, which revealed the following key insights:

为了解户外爱好者在自然探索与环境数据方面的需求，我开展了用户访谈，并从中得出以下关键洞察：

· Users desire specific nature data beyond basic weather.
· Nature education is most effective when integrated into routes via map markers or interactive tasks.
· Gamified collection of digital plant/animal cards (Pokémon-style) highly motivates exploration.
· Lightweight content (2–3 sentences + visuals) is preferred over long, cluttered information.

· 用户需要超越基础天气的具体自然数据。
· 将自然教育通过地图标记或互动任务融入路线中效果最佳。
· 游戏化的数字动植物卡片收集极大激励了探索热情。
· 用户偏好简短内容（2–3 句 + 视觉），而非冗长繁杂的信息。

As outdoor navigation and biodiversity identification tools remain fragmented, there is no single integrated solution to combine route planning with real-time ecological alerts and contextual species discovery, thereby restricting the quality of users' outdoor experience and their learning potential.

由于户外导航与生物多样性识别工具仍处于碎片化状态，目前尚无一体化解决方案能将路线规划、实时生态预警与情境化物种发现整合在一起，从而限制了用户户外体验的质量与学习潜力。

As outdoor navigation and biodiversity identification tools remain fragmented, there is no single integrated solution to combine route planning with real-time ecological alerts.

由于户外导航与生物多样性识别工具仍处于碎片化状态，目前尚无一体化方案能将路线规划与实时生态预警有机结合。

To validate diverse user needs, two distinct Customer Journey Maps were developed, building on the core user flow. These maps visualize user interactions across six key stages, immediately identifying their unique pain points and critical design opportunities.

为验证多样化的用户需求，在核心用户流程的基础上，我们开发了两份差异化的用户旅程地图。这些地图将用户在六个关键阶段的交互过程可视化，直观呈现各自的痛点与关键设计机会。

In summary, the CJMs reveal the necessary balance between structured learners and casual explorers.

综上所述，用户旅程地图揭示了系统学习者与休闲探索者之间所需的平衡。

The final IA was refined based on insights from mid-fidelity usability testing.

最终信息架构根据中保真可用性测试的洞察进行了优化。

Final usability testing validated the high-fidelity prototype, achieving a 100% Task Success Rate and a SUS score of 86.38/100.

The successful completion of this testing phase validates the core user flow, marking the effective conclusion of the MVP.

最终可用性测试验证了高保真原型，任务成功率达 100%，SUS 评分为 86.38/100。

测试阶段的顺利完成验证了核心用户流程，标志着最小可行产品（MVP）的有效交付。

Want the full process, from user research and ideation to two Lean UX cycles and usability testing? Read the complete design report.

想了解从用户调研、构思到两轮精益 UX 迭代与可用性测试的完整设计过程？查看完整设计报告。

Apply7 is an AI-powered platform that simplifies the international student study abroad process by offering university matching, document preparation, and application guidance.

I joined the Apply7 team as the UX Designer at its inception, spanning the full product cycle from building the design system and delivering the high-fidelity prototype for the key school matching feature to ensuring product maturity through cross-functional collaboration with the Hangzhou and San Diego teams.

Apply7 是一款 AI 驱动的平台，通过提供大学匹配、文件准备与申请指导，简化国际学生的海外留学申请流程。

我在 Apply7 创立之初以 UX 设计师的身份加入团队，参与了完整的产品周期——从构建设计系统、交付学校匹配核心功能的高保真原型，到与杭州和圣地亚哥团队跨职能协作，推动产品走向成熟。

The study abroad process is fragmented and complex, characterized by an information gap that causes significant stress for international students.

留学申请流程碎片化且复杂，信息不对称造成的信息差给国际学生带来了巨大压力。

International students preparing for overseas university applications.

正在准备申请海外大学的国际学生。

Apply7 is an integrated, AI-powered platform designed to close this information gap and simplify the process. Its core value lies in using AI to provide personalised school matching, assist with document revision and preparation, and generate a streamlined application plan.

Apply7 是一款一体化的 AI 驱动平台，旨在弥合信息差并简化申请流程。其核心价值在于利用 AI 提供个性化学校匹配、辅助文件修改与准备，并生成简洁高效的申请方案。

Working in an early-stage startup allowed me to take full ownership of the user experience. I was involved not only in designing interfaces, but also in shaping the overall product flow. This flow illustrates how I restructured the onboarding process to lower the entry barrier and enhance user engagement.

在早期初创公司工作，让我得以全面主导用户体验设计。我不仅参与了界面设计，还深度参与了整体产品流程的规划。以下流程图展示了我如何重构用户引导流程，以降低使用门槛、提升用户参与度。

We mapped out the core user flow and defined the primary structure of the application. I created mid-fidelity wireframes to explore layout and interactions, then transformed them into high-fidelity prototypes, starting with the school matching module. Throughout the process, I applied our design system and collaborated closely with the engineering team to ensure smooth handoff and implementation.

我们梳理了核心用户流程，并确定了应用的主要架构。我先通过中保真线框图探索布局与交互方式，再将其转化为高保真原型，从学校匹配模块开始推进。整个过程中，我持续运用设计系统，并与工程团队紧密协作，确保设计顺利交付与落地实现。

Visit Our Product: https://www.apply7.ai/

访问我们的产品：https://www.apply7.ai/

Gen Z is stuck in a cycle of emotional impulse spending, yet existing financial tools are ill-equipped to help. These apps often react to overspending after it happens, rather than helping users intervene in real time. This creates a persistent lack of reflection, perpetuating a cycle of regret and financial stress.

Z 世代陷入情绪化冲动消费的循环，而现有理财工具却难以提供有效帮助。这些应用往往在超支发生后才做出反应，而非帮助用户实时干预。这造成了持续缺乏反思的状态，使悔恨与财务压力的循环不断延续。

HMW transform the transaction-driven, reactive budgeting tool into an opportunities for mindful reflection and intentional financial decision-making?

我们如何将以交易为驱动的被动预算工具，转变为促进正念反思与有意识财务决策的机会？

Our target Gen Z aged 18–30 with impulsive spending habits who want to make long-term financial goals and build long-term financial literacy.

我们的目标用户是年龄在 18 至 30 岁之间、有冲动消费习惯的 Z 世代，他们希望制定长期财务目标并培养长期的财务素养。

By using the Lean UX method, Finspire was designed to introduce emotional check-ins, spending impact, and reflective nudges to favour mindful financial behaviour.

采用精益 UX 方法，Finspire 旨在引入情绪签到、消费影响提示与反思性推送，以促进理性的财务行为。

To understand the emotional and behavioural factors driving impulsive spending within Gen Z, I conducted in-depth interviews with individuals aged 18–30.

为了解驱动 Z 世代冲动消费的情绪与行为因素，我对 18 至 30 岁的用户进行了深度访谈。

Key Insights:
· Frictionless spending: mobile payments make spending feel effortless and invisible, so users spend on impulse without a second thought.
· Failed budgeting: 87.5% had tried budgeting apps, but 75% couldn't sustain the habit.
· Emotional buying: 75% of participants admitted to emotionally driven impulse purchases.
· Pure impulse: around 14% of purchases were made on a "just buy it" impulse with no real consideration.
· Desire for reflection: users want a moment to pause before buying, and often feel regret afterwards, hoping to better control their spending.

关键洞察：
· 无感消费：移动支付让花钱变得毫无门槛，用户往往不假思索地随手消费。
· 预算失败：87.5% 的用户曾尝试预算应用，但 75% 无法坚持使用。
· 情绪消费：75% 的参与者承认存在情绪驱动的冲动购买。
· 纯冲动购买：约 14% 的购买出于"买了再说"的冲动，缺乏真正的考量。
· 渴望反思：用户希望购买前能有片刻停顿，且常在购买后陷入懊悔，希望更好地掌控自己的消费。

Both primary and secondary research confirmed the core problem: Gen Z's impulse spending stems not from an irrational desire for money, but from deeply ingrained emotional spending patterns that current financial tools fail to address. Secondary research also revealed a clear market gap, as no existing product combines emotional and behavioural spending support with a data-backed, personalised approach. This defined Finspire's positioning: a product that intervenes in these emotional patterns, a direction that guided the entire design process.

一手与二手研究共同验证了核心问题：Z 世代的冲动消费并非源于对金钱的非理性渴望，而是根深蒂固的情绪消费模式，而现有理财工具未能解决这一问题。二手研究还发现了明显的市场空白，没有任何产品以数据支撑的个性化方式，同时提供情绪与行为层面的消费支持。这界定了 Finspire 的定位：一款介入用户情绪消费模式的产品，并以此方向指导整个设计过程。

To fully address the core problems, specifically targeting the design of the different form decision, impulsive purchase method and its key points. Finspire's system to ensure well to simulate and understand UX data.

为全面解决核心问题，特别针对不同形式的决策设计、冲动购买机制及其关键节点。Finspire 的系统旨在有效模拟并理解用户体验数据。

The Finspire style guide establishes the app's visual identity and interaction language, making it feel both trustworthy and approachable.

Finspire 风格指南确立了应用的视觉形象与交互语言，使其既值得信赖又平易近人。

‹ ›

Final usability testing validated the prototype, achieving a 100% Task Success Rate and a System Usability Scale of 76.00/100.

最终可用性测试验证了原型，任务成功率达 100%，系统可用性量表（SUS）评分为 76.00/100。

Want the full process, from literature review and user research to usability testing with quantified results? Read the complete design report.

想了解从文献综述、用户调研到量化可用性测试的完整设计过程？查看完整设计报告。

University students face the recurring challenge of moving every year — repeatedly buying, transporting, and disposing of furniture. Local students have limited budgets, while international students are completely unfamiliar with the local market, with language and cultural barriers making the problem even more complex.

大学生每年面临搬家，却要反复购买、运输、处置家具。本地生预算有限，国际学生则对市场完全陌生，语言和文化障碍让这个问题更加复杂。

"Flexible living, made simple."

"灵活居住，化繁为简。"

Let students access furniture at low cost and with flexibility — addressing pain points such as frequent moves, limited budgets, and international students' unfamiliarity with the local market.

让学生以低成本、灵活的方式获取家具，解决搬家频繁、预算有限、跨国学生不熟悉本地市场等痛点。

A complete UX process following the Double Diamond framework: in-depth user understanding through interviews and focus groups; construction of realistic personas; empathy maps and customer journey maps; ideation through HMW, Crazy 8's, and participatory design; NUF testing to filter the best solutions; and full iteration from paper prototype → clickable prototype → block wireframes → styled wireframes.

遵循双钻模型的完整 UX 流程：通过访谈和焦点小组深度理解用户；构建真实感强的用户画像；绘制同理心地图和客户旅程图；通过 HMW、Crazy 8's 和参与式设计产出方案；NUF 测试筛选最佳方向；从纸质原型→可点击原型→块状线框图→样式化线框图完整迭代。

Through one-on-one in-depth interviews and focus groups, I explored the real difficulties local and international students face across the entire furniture buying, transporting, and disposing process, while capturing collective behaviours and social influences such as housemates deciding together and community-based information sharing. The raw data was then extracted into theme-coded sticky notes, grouped into clusters, and digitalised into an affinity matrix, surfacing five core themes: Affordability, Convenience, Flexibility, Sustainability, and Trust.

通过一对一深度访谈和焦点小组，分别从本地生和国际生的视角理解家具购买、搬运、处置全过程中的真实困难，并捕捉室友共同决策、社群信息获取等集体行为与社会影响因素。随后将原始数据提取为按主题色区分的便利贴，归组识别核心主题簇，数字化整理成亲和图矩阵，最终归纳出五个核心主题：可负担性、便利性、灵活性、可持续性、信任感。

A lateral comparison of indirect competitors including IKEA, Airbnb, Spotify, TaskRabbit, and Notion across three dimensions: Similar Function, Similar User Behaviour, and Similar User Experience. The focus was not on finding direct competitors, but on learning how different industries solve the shared challenges of "flexibility" and "trust", and identifying interaction patterns worth borrowing.

Built separately for Callum (local undergraduate) and Lina (Chinese international postgraduate), systematically mapping user psychology across six dimensions — Says / Thinks / Does / Feels / Goals / Pain Points — to build a three-dimensional understanding of the full user picture.

分别为 Callum（本地本科生）和 Lina（中国国际研究生）构建，从 Says / Thinks / Does / Feels / Goals / Pain Points 六个维度系统梳理用户心理，建立对用户全貌的立体理解。

Two personas built from real research data: Callum, a 27-year-old local undergraduate at University of Bristol who shares a flat with housemates on a tight budget, and Lina, a 27-year-old Chinese international postgraduate studying Marketing who just arrived in the UK, facing both language and cultural barriers. For each, early and final journey maps were created across the complete five-stage journey (Awareness → Find → Pay → Receive → Return), annotating behaviours and touchpoints, thoughts and emotions, pain points, and design opportunities. The emotional curves reveal that payment is the biggest low point, while uncertainty at the return stage is a persistent source of anxiety.

基于真实研究数据构建两个用户画像：Callum，27岁布里斯托大学本地本科生，与室友合租，预算紧张；Lina，27岁中国国际研究生，Marketing硕士，刚到英国，语言和文化双重障碍。为两人分别绘制早期和最终版旅程图，覆盖 Awareness → Find → Pay → Receive → Return 完整五阶段，标注用户行为与接触点、想法与情绪、痛点和设计机会。情绪曲线显示：支付阶段是最大低谷，归还阶段的不确定感是持续焦虑来源。

Three key insights distilled from all research data:

基于所有研究数据提炼出3条最重要的洞察：

· International students' pain runs far deeper than locals': they lack moving experience, local market knowledge, and language ability, needing more guidance and assurance.
· Trust, not rent, is the core barrier to conversion: what students fear most is being charged extra on return.
· Bundles fit students' mental models better than browsing single items: they want to furnish a whole room, not pick every piece.

· 国际生的痛点远深于本地生：缺乏搬家经验、本地市场知识与语言能力，需要更多引导和保障。
· 信任，而非租金，是转化的核心障碍：学生最担心的是归还时被额外收费。
· 家具包比单件浏览更贴合学生心智：他们想要的是配齐一个房间，而不是逐件挑选。

HMW questions generated from POV insights, organised into 6 thematic clusters: Search & Browse · Price Transparency & Flexibility · Delivery & Installation · Return & Recycling · Customer Service & Maintenance · Community & Sustainability.

基于 POV 洞察生成 HMW 问题，分6个主题簇：搜索与浏览 · 价格透明度与灵活性 · 配送与安装 · 归还与循环利用 · 客服与维修支持 · 社区与可持续性。

Team members each sketched 8 interface concepts under time pressure, generating core directions: bundle browsing mode, student checklist feature, browsing by room type, AR try-out concept, and personalised recommendation homepage. Rapid ideation followed with more detailed single-page sketches and annotations for the most promising directions.

团队成员在限时内各自绘制8个界面概念草图，产出核心方向：家具包浏览模式、学生必备清单功能、按房间类型分类浏览、AR试摆功能概念、个性化推荐首页。随后进行快速创意生成，对最有潜力的方向做更详细的单页草图和注释。

I scored all ideas across three dimensions (New, Useful, and Feasible) to filter the feature set entering the prototype stage, then distilled the results into four design principles: design around students' real needs; make trust visible; guide rather than over-interrupt; and design for arrival, not just checkout.

对所有创意从 New（新颖度）、Useful（有用性）、Feasible（可行性）三个维度打分，筛选进入原型阶段的功能集，并据此提炼出四条设计原则：从学生的真实需求出发；让信任看得见；引导而非过度干扰；为到达而设计，而非仅为结账。

Starting from an app flow that mapped the full journey across key screens, I sketched hand-drawn lo-fi wireframes to explore layout and structure, then digitalised them into mid-fidelity wireframes. Together they cover the complete path from onboarding and browsing to cart, checkout, order tracking, and account.

先以 App Flow 梳理覆盖核心页面的完整流程，再通过手绘低保真线框探索布局与结构，最后数字化为中保真线框。三个阶段共同覆盖了从 Onboarding、浏览到购物车、结账、订单追踪与账户管理的完整操作路径。

Feedback testing on the clickable low-fidelity prototype; all participants completed their tasks and found the design helpful.

使用可点击低保真原型进行反馈测试，参与者均完成任务并认为设计有帮助。

Key findings and improvements:
· Post-checkout redirect should return users to the homepage, not My Orders.
· Product detail navigation was confusing and needs simplifying.
· Cart item count and subtotal hierarchy needs clearer structure.
· Delivery address form was missing a recipient name field.
· Minor copy and subtotal errors to be fixed.

核心发现与改进：
· 结账后应跳转回主页，而非 My Orders 页面。
· 商品详情页导航令人困惑，需简化。
· 购物车数量与小计的信息层级需更清晰。
· 收件地址栏缺少收件人姓名字段。
· 修正个别文案拼写与小计数字不匹配的问题。

Based on the insights gathered from feedback testing, the mid-fidelity wireframes were refined and elevated into a high-fidelity prototype, with improved visual design and interaction details.

基于反馈测试所获取的洞察，在中保真原型的基础上进行优化迭代，最终形成视觉设计与交互细节更为完善的高保真原型。

In this phase, I insisted on interviewing real users rather than generating user personas with AI. Through exploration and trial, I found that AI-generated interviews always stayed on the surface and couldn't surface the kind of deeply personal pain points that real users have. After the interviews, I handed the transcripts to AI for keyword extraction and synthesis, organising them into an affinity map in Miro; then, based on this real data, I built the empathy map and user journey map. Following the emotion curve, I identified the real problem this project needed to solve and defined the design opportunities.

这一阶段我坚持访谈真实用户，而没有用 AI 生成用户画像。经过探索和尝试，AI 生成的访谈总是浮在表面，提炼不出真实用户那种很个人化的痛点。访谈完成后，我把逐字稿交给 AI 做关键词提取和归纳，在 Miro 里整理成亲和图；再基于这些真实数据，搭出同理心地图和用户旅程图。顺着情绪曲线，我找到了这个项目真正要解决的问题，也定下了设计机会点。

During divergent thinking, AI was a prolific collaborator, like a tireless partner that could keep throwing out ideas, and it really sparked my thinking. That said, these "wild ideas" still needed my own reflection and filtering, keeping only the ones that truly held up.

在发散阶段，AI 是一个高产的协作伙伴，像一个永远不会累、能一直抛想法的搭档，很能启发我。当然，这些“脑洞”也需要我自己再思考和筛选，留下真正站得住的部分。

I first drew the app flow and wireframes, but found that without a clear information hierarchy, Claude struggled to output work with a coherent design system. So I first completed a more refined mid-fidelity prototype, distilled an initial design system, and after adjusting the framework through user testing, handed it to Claude Design to produce the high-fidelity solution. This time, the output quality improved significantly, reaching a state that was essentially ready for development.

我先绘制了 App Flow 与线框图，但发现当缺少清晰的信息层级时，Claude 难以输出具备设计系统的方案。因此我先完成更精细的中保真原型，归纳出初步的设计系统，经用户测试调整框架后，再交由 Claude Design 输出高保真方案。这一次，产出质量显著提升，基本达到可直接开发的程度。

This project explored a Figma + Claude workflow, and my biggest takeaway is this: the greatest value of AI in prototyping lies in its extremely fast iteration speed. It can quickly visualise ideas and help me filter out the best solution from many options. But the prerequisite for all of this is that the designer must first establish a clear information architecture and an initial design system, setting actionable boundaries for the AI. AI handles the speed, while defining the problem, making trade-offs, and taking responsibility for the final decision remain the designer's irreplaceable core value.

本次项目探索了 Figma + Claude 的协作方式，最大的体会是：AI 参与原型设计的最大价值，在于极高的迭代速度，它能快速将想法可视化，帮助我在多个方案中筛选出最优解。但这一切的前提是，设计师需要先建立清晰的信息架构与初步的设计系统，为 AI 划定可执行的边界。AI 负责提速，而定义问题、做出取舍、为最终决策负责，始终是设计师不可替代的核心价值。

Using AI tools in group chats requires switching between different apps, which disrupts the collaboration flow and breaks conversational context.

在群聊中使用 AI 工具需要在不同应用间来回切换，打断协作流程并丢失对话上下文。

How might we seamlessly integrate ChatGPT into group chat interfaces while keeping the experience simple and intuitive?

如何将 ChatGPT 无缝整合进群聊界面，同时保持操作的简洁性与直觉性？

Users who rely on group chats for collaboration, such as students and work teams.

依赖群聊进行协作的用户，如学生与工作团队。

A two-column layout that allows users to summon ChatGPT directly in group chats via @ChatGPT, while also supporting private ChatGPT conversations for personalised queries — all without leaving the app.

采用两栏布局，用户可在群聊中通过 @ChatGPT 直接调用 AI，同时支持私聊 ChatGPT 处理个人需求，全程无需离开应用。

The study examines messaging apps and AI tools across three levels: layout, features, and user behaviour. Layouts are categorised into Two-Column Interfaces (e.g., WhatsApp, Slack) suited for simple or mobile scenarios, and Three-Column Interfaces (e.g., Discord) for complex interactions. Behavioural analysis highlights that high-frequency features like text and voice dominate daily use, while low-frequency ones like translation address task-specific needs — underscoring the importance of prioritising primary operations.

本研究从布局、功能与用户行为三个层面对即时通讯应用与 AI 工具进行分析。布局分为两栏界面（如 WhatsApp、Slack），适用于简单或移动端场景；以及三栏界面（如 Discord），适用于复杂交互。行为分析表明，文字、语音等高频功能主导日常使用，而翻译等低频功能则用于特定任务，由此强调了优先保障核心操作的重要性。

The initial design adopted a two-column layout, allowing ChatGPT to be summoned via "@ChatGPT" in group chats. To handle complex queries, a private chat window was introduced in a three-column layout. However, the crowded interface and limited mobile compatibility led to a refined two-column design that merges group and private chat functions for better efficiency and usability.

初始设计采用两栏布局，允许用户在群聊中通过 "@ChatGPT" 召唤 AI。为处理复杂查询，引入了三栏布局的私聊窗口。然而，界面过于拥挤且移动端兼容性有限，最终优化为融合群聊与私聊功能的两栏设计，提升了效率与可用性。

The interface uses a two-column layout, holding the group chat and the private ChatGPT window separately, with each screen focused on one primary action to reduce visual clutter. A restrained graphite-grey neutral palette forms the base, creating a calm, professional feel; purple appears only as an accent, lighting up key interactions such as the send button, @ChatGPT mentions, and selected states, signalling innovation and the presence of AI. ChatGPT's replies use a distinct style: light-toned bubbles with a dedicated name label, clearly set apart from user messages, while keeping the original quoted question and timestamp to improve clarity and traceability.

界面采用两栏布局，分别承载群聊与 ChatGPT 私聊窗口，每屏聚焦一项主要操作，减少视觉干扰。整体以克制的石墨灰中性色调为基底，营造冷静、专业的氛围；紫色仅作点缀，在发送按钮、@ChatGPT 提及与选中态等关键交互处点亮，象征创新与 AI 的介入。ChatGPT 的回复采用独立样式，浅色气泡搭配专属名称标识，与用户消息清晰区分，并保留所引用的原始问题与时间戳，增强对话的清晰度与可追溯性。

Users can ask ChatGPT by right-clicking a message to reply and @ChatGPT, or by directly typing @ChatGPT in the input box. This design maintains human control while allowing flexible, explicit question customisation aligned with real-life conversational patterns.

用户可通过右键消息选择回复并 @ChatGPT，或直接在输入框中输入 @ChatGPT 发起提问。该设计在保持人工控制的同时，允许灵活的个性化提问，符合现实对话习惯。

Users can send messages to ChatGPT privately by right-clicking and selecting "Send to ChatGPT privately." The system offers options for an existing or new private chat window and automatically transitions to the selected window, ensuring efficient task completion.

用户可通过右键选择"私发给 ChatGPT"，系统提供选择已有或新建私聊窗口的选项，并自动跳转至所选窗口，确保任务高效完成。

To address miscommunication between people, ChatGPT detects ambiguous messages in real-time and prompts users to clarify via a "suggested revision" above the input box. For ambiguous queries directed at ChatGPT, the AI first expresses uncertainty, infers user intent, and then provides a response — reducing friction on both sides of the conversation.

针对人与人之间的误解，ChatGPT 实时检测模糊信息，并通过输入框上方的"建议修改"提示用户澄清。对于用户向 ChatGPT 提出的模糊问题，AI 会先表达不确定性，推断用户意图后再给出回应，有效降低双向沟通的摩擦。