[{"data":1,"prerenderedAt":578},["ShallowReactive",2],{"morph-inspect":3,"work-nav-all":529},{"id":4,"title":5,"body":7,"cover":480,"description":454,"details":481,"extension":503,"headline":504,"meta":507,"navigation":508,"order":509,"path":510,"role":511,"seo":512,"seoTitle":514,"slug":517,"status":518,"stem":517,"summary":519,"tags":522,"year":527,"__hash__":528},"works\u002Fmorph-inspect.md",{"zh":6,"en":6},"Morph Inspect",{"type":8,"value":9,"toc":453},"minimark",[10,261],[11,12,13,18,22,43,47,50,53,56,65,73,81,85,88,96,99,103,106,109,115,118,121,124,130,133,136,139,145,149,152,155,161,165,168,171,177,181,184,213,218,227,231,234],"zh",{},[14,15,17],"h2",{"id":16},"一場-2-小時的房屋檢查回到辦公室後還要再花-23-倍時間寫報告","一場 2 小時的房屋檢查，回到辦公室後還要再花 2–3 倍時間寫報告",[19,20,21],"p",{},"在美國的房產交易中，房屋檢查 (Home Inspection) 是對房屋現況進行全面檢視。房屋檢查員 (Home Inspector) 會記錄當下屋況，將現場觀察整理成一份詳實報告，作為確保交易公平與安全的重要依據。",[23,24,27,36],"grid",{"align":25,"cols":26},"center",2,[28,29,30,33],"cell",{},[19,31,32],{},"一場現場檢查可能約 2 小時，但檢查員回到辦公室後，往往還要再花 2–3 倍時間整理照片、歸類項目、撰寫屋況描述。",[19,34,35],{},"我們訪談了 10 多位房屋檢查員，發現他們的痛點不是不會寫，而是報告的整理佔掉太多時間與精力。",[28,37,38],{},[39,40],"inline-video",{"label":41,"src":42},"Home inspection workflow showing the gap between fieldwork and report writing","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Fworkflow-report-gap.mp4",[14,44,46],{"id":45},"房屋檢查報告須中立客觀ai-不能替檢查員做專業判斷","房屋檢查報告須中立客觀，AI 不能替檢查員做專業判斷",[19,48,49],{},"房屋檢查報告著重「發現問題」而非「給予保證」，報告必須保持中立客觀，避免過度推估成因、損害預測、或主觀評價。即便 AI 生成能力強大，也不能取代檢查員的專業判斷。",[19,51,52],{},"同時，房屋檢查產業較為傳統，檢查員的工具與工作習慣根深蒂固、切換成本高，我們更需要讓 AI 以「融入」而非「顛覆」的方式提升效率。",[19,54,55],{},"這些界線是用來建立檢查員對 AI 的信任感，讓檢查員敢把它用在真實工作裡。",[57,58,62],"info-card",{"icon":59,"title-en":60,"title-zh":61},"material-symbols:gavel","AI Defers to Expertise","AI 不做專業判斷",[19,63,64],{},"損害嚴重性由檢查員判斷，AI 只負責輔助整理與生成。",[57,66,70],{"icon":67,"title-en":68,"title-zh":69},"material-symbols:tune","Controllable Output","生成內容必須可控",[19,71,72],{},"檢查員能快速檢視、編輯與修正所有 AI 生成的內容。",[57,74,78],{"icon":75,"title-en":76,"title-zh":77},"material-symbols:merge","Fit Into Existing Workflows","AI 需融入既有流程",[19,79,80],{},"不打斷現場節奏，並降低工具切換成本。",[14,82,84],{"id":83},"把-ai-當成辦公室秘書不打斷現場節奏但在背後把報告整理好","把 AI 當成辦公室秘書，不打斷現場節奏，但在背後把報告整理好",[19,86,87],{},"我們把 AI 定位成一位「辦公室裡的秘書」，而不是單純的聊天機器人。檢查員照原本節奏在現場拍照與簡短紀錄，AI 在背後接住素材、整理成可用的報告草稿。換句話說，檢查完成時，報告也已接近完成。",[89,90],"full-image",{"alt":91,"caption-en":92,"caption-zh":93,"src":94,":lightbox":95},"AI 擔任辦公室秘書，在背後整理資料讓報告在檢查完成時同步就緒","AI acts as the back-office secretary. When inspection ends, the report is nearly done.","AI 作為背後的秘書，讓檢查完成時報告也近乎完成","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Fai-office-secretary.webp","false",[14,97,98],{"id":98},"三大核心功能",[100,101,102],"h3",{"id":102},"拍照先行的流程",[19,104,105],{},"我們觀察到檢查員在現場需要快速移動，在有限時間內走遍房屋各個角落，當發現問題時，第一個動作就是拍照記錄。",[19,107,108],{},"因此流程設計以拍照作為起點，並可視情況增加文字筆記或語音轉錄作為補充，讓紀錄方式符合現場檢查工作節奏。",[89,110],{"alt":111,"caption-en":112,"caption-zh":113,"src":114},"先拍照的工作流程：現場先拍照，再補充文字筆記","Capture photos first, then add notes without breaking fieldwork.","先拍照再補文字，讓現場紀錄不中斷。","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Ffeature-photo-first.webp",[100,116,117],{"id":117},"生成可控",[19,119,120],{},"損害嚴重性 (concerning level) 由檢查員選定，AI 會依照片＋標註＋筆記 生成描述，檢查員能快速檢視與編修。",[19,122,123],{},"設計核心在於「不讓 AI 自行判定損害嚴重性」，避免 AI 過度推估、生成與現場實況不符的描述，確保報告描述始終基於檢查員的專業判斷。",[89,125],{"alt":126,"caption-en":127,"caption-zh":128,"src":129},"生成可控：檢查員選擇嚴重程度，AI 產生描述草稿","Inspectors choose severity; AI drafts a controlled description.","檢查員選擇嚴重程度，AI 產生可控的描述草稿。","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Ffeature-controllable-generation.webp",[100,131,132],{"id":132},"自動歸類",[19,134,135],{},"檢查員在現場的移動路徑，並不會照著報告的章節順序走，可能先進地下室、再上二樓、最後才到屋外。",[19,137,138],{},"我們設計讓 AI 自動將各處紀錄歸入對應的報告章節，省去事後手動整理的步驟，讓檢查員在現場只需專注於發現問題，不必分心想「這條記錄要放哪裡」。",[89,140],{"alt":141,"caption-en":142,"caption-zh":143,"src":144},"自動歸類：AI 將缺失項目自動分類到報告各章節","AI categorizes findings automatically, reducing report cleanup time.","AI 自動歸類缺失項目，減少整理報告的時間。","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Ffeature-automatic-categorization.webp",[14,146,148],{"id":147},"不用維護模板庫用舊報告與編輯回饋自然長出個人風格","不用維護模板庫，用舊報告與編輯回饋，自然長出個人風格",[19,150,151],{},"傳統報告工具多半靠模板庫加速產出，檢查員勾選描述再微調文字。但模板庫需要長期維護，越用越大，最後反而變成負擔。",[19,153,154],{},"我們將「維護模板」轉變成「自然累積」的方式，檢查員上傳報告作為參考資料，AI 從中檢索出風格相近的描述片段，再結合當下的照片與筆記生成內容 (RAG)。每次人工編修也會回饋到系統，讓輸出逐步貼近個人風格，但不需額外心力維護。",[89,156],{"alt":157,"caption-en":158,"caption-zh":159,"src":160},"上傳舊報告作為參考資料、RAG 檢索，以及個人風格的自然累積","From uploading past reports to RAG retrieval, to natural personal style accumulation","從上傳舊報告到 RAG 檢索，再到個人風格自然累積","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Frag-past-reports.webp",[14,162,164],{"id":163},"延伸設計ai-預填欄位減少缺漏與重複確認","延伸設計：AI 預填欄位，減少缺漏與重複確認",[19,166,167],{},"有些州\u002F地區對房屋檢查報告有指定格式與必填欄位，若現場檢查時漏填，事後往往得再花時間回頭確認補齊。",[19,169,170],{},"基於 AI 已能理解紀錄內容並完成整理，我提出讓 AI 在整理內容後先預填寫欄位、再由檢查員確認的概念，讓資料補齊跟著流程一起完成，也讓 AI 更進一步自然融入檢查流程。（concept）",[89,172],{"alt":173,"caption-en":174,"caption-zh":175,"src":176},"AI 預填必填欄位，檢查員確認","Pre-fill then confirm: AI pre-fills mandatory fields based on field records. Inspectors only need to confirm or correct, with no more hunting for missing data after the fact.","預填後確認：AI 先根據現場紀錄預填必填欄位，檢查員只需確認或修正，不再需要事後回頭補資料。","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Ffeature-prefill-confirm.webp",[14,178,180],{"id":179},"成果產品落地驗證獲得展會好評與合作機會","成果：產品落地驗證，獲得展會好評與合作機會",[19,182,183],{},"團隊在全員兼職、跨時區遠端的條件下，以小步快跑與異步協作推進開發，並把產品帶到市場驗證，獲得正面回饋。",[185,186,187,195,201,207],"ul",{},[188,189,190,194],"li",{},[191,192,193],"strong",{},"2025 Q2","：V1 正式上線",[188,196,197,200],{},[191,198,199],{},"2025 Q4","：參加 InterNACHI convention 展會，獲得 30+ 位檢查員的高度關注",[188,202,203,206],{},[191,204,205],{},"2026 Q1","：與美國地方政府機構接洽，獲得合作機會",[188,208,209,212],{},[191,210,211],{},"未來規劃","：離線模式、法規檢索、CRM、Web 報告編修等功能開發",[89,214],{"alt":215,"caption-en":216,"caption-zh":216,"src":217,":lightbox":95},"團隊於 TPREIA-InterNACHI Texas Inspectors Convention 2025","TPREIA-InterNACHI® Texas Inspectors Convention, 2025","\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Ftexas-inspectors-convention-2025.webp",[219,220],"video-block",{"aspect":221,"caption-en":222,"caption-zh":223,"src":224,"title":225,"type":226},"portrait","Morph Inspect promotional video","Morph Inspect 宣傳短片","https:\u002F\u002Fplayer.vimeo.com\u002Fvideo\u002F1189409286?title=0&byline=0&portrait=0&badge=0&dnt=1&transparent=0","Morph Demo","iframe",[14,228,230],{"id":229},"反思重新理解-ai-在專業工具中的角色","反思：重新理解 AI 在專業工具中的角色",[19,232,233],{},"一開始我認為 AI 會用新穎且顛覆性的方式改變房屋檢查流程，但真正能落地的價值，反而是 AI 不取代檢查員的專業，在最低打擾的前提下融入既有流程，讓檢查員更專注在檢查工作與提升效率。",[235,236,238,246,253],"info-cards",{"cols":237},3,[57,239,243],{"icon":240,"title-en":241,"title-zh":242},"material-symbols:engineering","Amplify, Not Replace","AI 不是取代專業，而是放大專業",[19,244,245],{},"判斷仍由檢查員掌握，AI 接手整理麻煩的任務。",[57,247,250],{"icon":75,"title-en":248,"title-zh":249},"Integrate, Don't Disrupt","融入而非顛覆",[19,251,252],{},"讓檢查員維持既有流程，效率自然提升，而不是額外學習新工具。",[57,254,258],{"icon":255,"title-en":256,"title-zh":257},"material-symbols:verified-user","Trust First","信任感是前提",[19,259,260],{},"生成內容可控、可修正，建立信任感，檢查員才會接納 AI。",[262,263,264,268,271,285,289,292,295,298,315,319,322,325,329,333,336,339,342,346,349,352,355,359,362,365,368,372,375,378,381,385,388,391,394,398,401,424,427,429,433,436],"en",{},[14,265,267],{"id":266},"a-2-hour-inspection-then-23-that-time-writing-the-report","A 2-Hour Inspection, Then 2–3× That Time Writing the Report",[19,269,270],{},"In U.S. real estate transactions, home inspection provides comprehensive assessment of property conditions. Home inspectors document current conditions and organize field observations into detailed reports, serving as critical documentation for fair and safe transactions.",[23,272,273,281],{"align":25,"cols":26},[28,274,275,278],{},[19,276,277],{},"While on-site inspection takes approximately 2 hours, inspectors often spend 2–3 times longer organizing photos, categorizing items, and writing property condition descriptions back at the office.",[19,279,280],{},"After interviewing over 10 home inspectors, we discovered their challenge isn't inability to write. It's organization consuming excessive time and effort.",[28,282,283],{},[39,284],{"label":41,"src":42},[14,286,288],{"id":287},"reports-must-be-neutral-ai-cant-make-professional-judgments","Reports Must Be Neutral. AI Can't Make Professional Judgments",[19,290,291],{},"Inspection reports emphasize \"problem discovery\" rather than \"providing guarantees.\" They must remain neutral and objective, avoiding overstated causes, damage predictions, or subjective assessments. Despite AI's strong generation capabilities, it cannot replace inspector expertise.",[19,293,294],{},"The home inspection industry is traditionally entrenched, with established tool use and work habits carrying high switching costs. We needed AI to enhance efficiency through integration rather than disruption.",[19,296,297],{},"These boundaries establish inspector confidence in AI, enabling real-world workplace adoption.",[235,299,300,305,310],{"cols":237},[57,301,302],{"icon":59,"title-en":60,"title-zh":61},[19,303,304],{},"Damage severity assessment remains the inspector's call. AI assists with organization and generation only.",[57,306,307],{"icon":67,"title-en":68,"title-zh":69},[19,308,309],{},"Inspectors can quickly review, edit, and correct all AI-generated content.",[57,311,312],{"icon":75,"title-en":76,"title-zh":77},[19,313,314],{},"No disruption to field pace, and no added tool-switching overhead.",[14,316,318],{"id":317},"position-ai-as-an-office-secretary-organizing-in-the-background","Position AI as an Office Secretary, Organizing in the Background",[19,320,321],{},"We positioned AI as an \"office secretary\" rather than a basic chatbot. Inspectors photograph and briefly note findings at their natural pace; AI captures those materials and organizes them into a usable report draft. In other words, when field inspection concludes, the report is nearly complete.",[89,323],{"alt":324,"caption-en":92,"caption-zh":93,"src":94,":lightbox":95},"AI positioned as office secretary, organizing materials in the background while inspector works in the field",[14,326,328],{"id":327},"three-core-features","Three Core Features",[100,330,332],{"id":331},"photo-first-workflow","Photo-First Workflow",[19,334,335],{},"Inspectors need to move quickly through a property, covering every corner in limited time. When something catches their eye, the first instinct is to reach for the phone and take a photo.",[19,337,338],{},"So we designed the flow around photography as the entry point, with text notes or voice transcription as optional supplements rather than required steps, keeping the recording process in step with how inspectors actually work on-site.",[89,340],{"alt":341,"caption-en":112,"caption-zh":113,"src":114},"Photo-first workflow: inspector captures photos before adding notes",[100,343,345],{"id":344},"controllable-generation","Controllable Generation",[19,347,348],{},"Inspectors select the damage severity level; AI generates a description based on photos, annotations, and notes. Inspectors can quickly review and edit the result.",[19,350,351],{},"The core design principle is that AI never determines severity on its own, preventing overestimation and descriptions that don't reflect field conditions, so that report content always rests on the inspector's professional judgment.",[89,353],{"alt":354,"caption-en":127,"caption-zh":128,"src":129},"Controllable generation: inspector selects severity, AI generates description",[100,356,358],{"id":357},"automatic-categorization","Automatic Categorization",[19,360,361],{},"An inspector's path through a property doesn't follow the report's chapter order. They might start in the basement, head upstairs, then finish outside.",[19,363,364],{},"AI automatically files each item into the corresponding report section, eliminating post-inspection sorting. Inspectors can stay focused on finding issues rather than thinking about where each note belongs.",[89,366],{"alt":367,"caption-en":142,"caption-zh":143,"src":144},"Automatic categorization: AI organizes items into report sections",[14,369,371],{"id":370},"no-template-library-personal-style-builds-itself","No Template Library. Personal Style Builds Itself",[19,373,374],{},"Traditional reporting tools rely on template libraries. Inspectors select descriptions and fine-tune text, but libraries require ongoing maintenance, grow unwieldy over time, and end up becoming a burden.",[19,376,377],{},"We shifted from \"template maintenance\" to \"natural accumulation.\" Inspectors upload past reports as reference material; AI retrieves stylistically similar passages from them and combines those with current photos and notes to generate descriptions (RAG). Each manual edit feeds back into the system, gradually shaping output toward personal style, without any extra maintenance effort.",[89,379],{"alt":380,"caption-en":158,"caption-zh":159,"src":160},"Past reports uploaded as reference material, RAG retrieval, and personal style accumulation",[14,382,384],{"id":383},"extension-design-ai-pre-fills-fields-to-reduce-omissions","Extension Design: AI Pre-Fills Fields to Reduce Omissions",[19,386,387],{},"Some states and regions require specific report formats with mandatory fields. Omissions during field inspection often mean additional time spent confirming and completing data afterward.",[19,389,390],{},"Since AI already understands record content and completes organization, I proposed having AI pre-fill mandatory fields after organizing content, with inspector confirmation. This allows data completion as part of the natural workflow, integrating AI even more seamlessly into the inspection process. (concept)",[89,392],{"alt":393,"caption-en":174,"caption-zh":175,"src":176},"AI pre-fills mandatory fields, inspector confirms",[14,395,397],{"id":396},"results-market-validation-trade-show-traction-partnership-opportunities","Results: Market Validation, Trade Show Traction, Partnership Opportunities",[19,399,400],{},"Operating with a part-time team across time zones, we used rapid iteration and asynchronous collaboration to advance development and bring the product to market for validation, earning positive feedback.",[185,402,403,408,413,418],{},[188,404,405,407],{},[191,406,193],{},": V1 official launch",[188,409,410,412],{},[191,411,199],{},": Attended InterNACHI convention, received attention from 30+ inspectors",[188,414,415,417],{},[191,416,205],{},": Began partnering with U.S. local government agencies",[188,419,420,423],{},[191,421,422],{},"Roadmap",": Offline mode, regulation lookup, CRM integration, web-based report editing, and more",[89,425],{"alt":426,"caption-en":216,"caption-zh":216,"src":217,":lightbox":95},"Team at TPREIA-InterNACHI Texas Inspectors Convention 2025",[219,428],{"aspect":221,"caption-en":222,"caption-zh":223,"src":224,"title":225,"type":226},[14,430,432],{"id":431},"reflection-rethinking-ais-role-in-professional-tools","Reflection: Rethinking AI's Role in Professional Tools",[19,434,435],{},"Initially, I believed AI would transform home inspection through novel, disruptive approaches. True value instead came from AI not replacing inspector expertise, but integrating into existing workflows with minimal disruption, allowing inspectors to focus on their work and gain real efficiency.",[235,437,438,443,448],{"cols":237},[57,439,440],{"icon":240,"title-en":241,"title-zh":242},[19,441,442],{},"Inspectors retain judgment authority while AI handles the tedious organization work.",[57,444,445],{"icon":75,"title-en":248,"title-zh":249},[19,446,447],{},"Inspectors maintain established workflows. Efficiency gains emerge naturally, with no new tool to learn.",[57,449,450],{"icon":255,"title-en":256,"title-zh":257},[19,451,452],{},"Controllable, correctable generated content builds the confidence inspectors need to adopt AI in real work.",{"title":454,"searchDepth":26,"depth":26,"links":455},"",[456,457,458,459,464,465,466,467,468,469,470,471,476,477,478,479],{"id":16,"depth":26,"text":17},{"id":45,"depth":26,"text":46},{"id":83,"depth":26,"text":84},{"id":98,"depth":26,"text":98,"children":460},[461,462,463],{"id":102,"depth":237,"text":102},{"id":117,"depth":237,"text":117},{"id":132,"depth":237,"text":132},{"id":147,"depth":26,"text":148},{"id":163,"depth":26,"text":164},{"id":179,"depth":26,"text":180},{"id":229,"depth":26,"text":230},{"id":266,"depth":26,"text":267},{"id":287,"depth":26,"text":288},{"id":317,"depth":26,"text":318},{"id":327,"depth":26,"text":328,"children":472},[473,474,475],{"id":331,"depth":237,"text":332},{"id":344,"depth":237,"text":345},{"id":357,"depth":237,"text":358},{"id":370,"depth":26,"text":371},{"id":383,"depth":26,"text":384},{"id":396,"depth":26,"text":397},{"id":431,"depth":26,"text":432},"\u002Fimages\u002Fworks\u002Fmorph-inspect\u002Fcover.webp",[482,489,496],{"label":483,"value":486},{"zh":484,"en":485},"角色","Role",{"zh":487,"en":488},"UI\u002FUX 設計師","UI\u002FUX Designer",{"label":490,"value":493},{"zh":491,"en":492},"團隊","Team",{"zh":494,"en":495},"2 位設計師、1 位 PM、4 位工程師（跨時區協作）","2 designers, 1 PM, 4 engineers (cross-timezone collaboration)",{"label":497,"value":500,"href":502},{"zh":498,"en":499},"產品","Product",{"zh":501,"en":501},"AI-Powered Inspection - Morph","https:\u002F\u002Fmorph-technologies.com\u002F","md",{"zh":505,"en":506},"將 AI 融入既有房屋檢查流程，協助檢查員有效縮短檢查報告撰寫時間","Integrating AI into existing home inspection workflows to help inspectors reduce report writing time",{},true,1,"\u002Fmorph-inspect",{"zh":487,"en":488},{"title":513,"description":454},{"zh":6,"en":6},{"zh":515,"en":516},"Morph Inspect - AI 輔助房屋檢查報告","Morph Inspect: AI-assisted home inspection reports","morph-inspect","published",{"zh":520,"en":521},"透過「人為判斷、AI 整理」的協作機制，讓檢查員在不改變既有工作流程的前提下，大幅縮減報告撰寫時間。以融入而非顛覆的設計，保留了專業判斷的主導權，建立了檢查員對 AI 工具的信任感。","Through a collaborative model of human judgment and AI organization, inspectors can significantly reduce report writing time without changing their existing workflow. By integrating rather than disrupting, the design preserves professional control and builds trust in AI tools.",[523,524,525,526],"AI","B2B","Mobile","Web",2025,"TseK0eVRSptMez60zSLYZ_Ufq3uJHLIcUZwJqSiLIlA",[530,533,542,551,560,568],{"slug":517,"title":531,"cover":480,"tags":532,"order":509},{"zh":6,"en":6},[523,524,525,526],{"slug":534,"title":535,"cover":537,"tags":538,"order":26},"spin-the-wheel",{"zh":536,"en":536},"Spin The Wheel","\u002Fimages\u002Fworks\u002Fspin-the-wheel\u002Fcover.webp",[539,540,541],"Wearable","Connect IQ","Dev",{"slug":543,"title":544,"cover":547,"tags":548,"order":237},"sim-monitor-dashboard",{"zh":545,"en":546},"散熱模擬監控儀表板","Thermal Simulation Monitoring Dashboard","\u002Fimages\u002Fworks\u002Fsim-monitor-dashboard\u002Fcover.webp",[549,550,541],"Design Challenge","IA",{"slug":552,"title":553,"cover":556,"tags":557,"order":559},"tiger-master",{"zh":554,"en":555},"師虎來了(修繕媒合平台)","TigerMaster Repair Matching Platform","\u002Fimages\u002Fworks\u002Ftiger-master\u002Fcover.webp",[558,525,541],"B2C",4,{"slug":561,"title":562,"cover":564,"tags":565,"order":567},"yodewang-rebuild",{"zh":563,"en":563},"yodewang.com","\u002Fimages\u002Fworks\u002Fyodewang-rebuild\u002Fcover.webp",[566,523,541],"Portfolio",5,{"slug":569,"title":570,"cover":573,"tags":574,"order":577},"ai-design-system-assistant",{"zh":571,"en":572},"建立一套讓 AI 看懂設計脈絡的知識庫","Building a Knowledge Base That Teaches AI Design Context","\u002Fimages\u002Fworks\u002Fai-design-system-assistant\u002Fcover.webp",[575,576,523],"[WIP]","Design System",6,1784180137454]