{ "schemaVersion": 1, "version": "2026-09-15", "id": "https://hk.onyxdevslab.com/data/enterprise-ai-data-readiness.json", "name": "Enterprise AI data-readiness checklist", "datePublished": "2026-09-15", "dateModified": "2026-09-15", "evidenceClass": "Provider-authored implementation checklist", "publisher": { "name": "Onyx Devs Lab", "legalName": "ONYX DEVS LAB LIMITED", "url": "https://hk.onyxdevslab.com/", "leiCode": "254900Z30CLK7HKE9H46" }, "landingPages": { "en": "https://hk.onyxdevslab.com/en/guides/enterprise-ai-data-readiness/", "zhHant": "https://hk.onyxdevslab.com/zh-hk/guides/enterprise-ai-data-readiness/", "zhHans": "https://hk.onyxdevslab.com/zh-cn/guides/enterprise-ai-data-readiness/" }, "languages": { "en": { "name": "How can an enterprise tell whether its data is ready for AI?", "summary": "Data readiness is task-specific. The enterprise does not need perfectly clean data everywhere; it needs the right records, meanings, permissions, history, evidence labels, and feedback for one defined workflow and failure consequence.", "decisionSignals": [ { "name": "Relevant", "meaning": "Connected to a defined decision and outcome" }, { "name": "Trustworthy", "meaning": "Owned, traceable, interpretable, and permissioned" }, { "name": "Usable", "meaning": "Representative, accessible, testable, and maintainable" } ], "sections": [ { "heading": "Evaluate data in the context of one workflow", "introduction": "A warehouse inventory or row count does not establish readiness for a particular AI decision.", "checks": [ { "name": "Source and ownership", "requirement": "Name systems of record, business and technical owners, collection purpose, authority to use, retention, residency, exports, and deletion obligations." }, { "name": "Meaning and identity", "requirement": "Define objects, keys, timestamps, units, status, versions, relationships, corrections, missingness, and how the same entity is reconciled across systems." }, { "name": "Representation and ground truth", "requirement": "Check coverage across users, periods, languages, common and rare cases, outcomes, labels, disagreement, selection bias, leakage, and changes over time." } ] }, { "heading": "Prove that data can support production", "introduction": "Readiness includes the pipeline and controls that keep evidence usable after the initial export.", "checks": [ { "name": "Access and protection", "requirement": "Enforce tenant, role, object, field, purpose, environment, and time boundaries; mask or minimise sensitive data and record access." }, { "name": "Pipeline and retrieval", "requirement": "Test ingestion completeness, parsing, duplicates, late and failed records, lineage, indexes, retrieval recall, citations, freshness, and reprocessing." }, { "name": "Feedback and maintenance", "requirement": "Capture corrections, exceptions, outcomes, drift, policy changes, ownership, service levels, incident handling, and the cost of keeping data useful." } ] } ], "procedure": [ { "position": 1, "name": "Define the decision", "instruction": "Specify user, workflow, output, system of record, failure consequence, freshness, evidence, and action boundary." }, { "position": 2, "name": "Create a source map", "instruction": "Inventory required fields, meaning, owner, permission, history, quality, access method, and known gaps—not all enterprise data." }, { "position": 3, "name": "Test a representative sample", "instruction": "Run ingestion, retrieval or modelling on ordinary, edge, missing, stale, conflicting, and restricted cases; record unknowns." }, { "position": 4, "name": "Choose the smallest remediation", "instruction": "Fix only gaps that block the decision: identity, collection, labelling, access, quality rules, feedback, or an alternative workflow." } ], "frequentlyAskedQuestions": [ { "question": "Must all data be cleaned before starting?", "answer": "No. Start from one valuable workflow and make the required evidence fit for that decision. Enterprise-wide cleanup without a decision boundary often wastes effort." }, { "question": "Is more historical data always better?", "answer": "No. Old data may reflect obsolete policy, bias, products, users, or processes. Relevance, coverage, rights, labels, and known change matter more than volume alone." }, { "question": "Can synthetic data solve missing data?", "answer": "It can support testing or rare-case coverage when clearly labelled, but it cannot prove real-world prevalence, behaviour, rights, or outcomes without validation." } ] }, "zhHant": { "name": "企業如何判斷數據是否已適合做 AI？", "summary": "數據就緒必須按任務判斷。企業不需要先把所有數據清洗至完美；它需要為一條已定義流程及失敗後果準備正確記錄、語義、權限、歷史、證據標註及反饋。", "decisionSignals": [ { "name": "相關", "meaning": "連接已定義決策及成果" }, { "name": "可信", "meaning": "有人負責、可追溯、可解讀、有權限" }, { "name": "可用", "meaning": "具代表性、可存取、可測試、可維護" } ], "sections": [ { "heading": "在一條流程中評估數據", "introduction": "數據倉庫清單或行數不能證明某項 AI 決策已就緒。", "checks": [ { "name": "來源及責任", "requirement": "列明記錄系統、業務與技術負責人、收集目的、使用權、保留、地區、導出及刪除責任。" }, { "name": "語義及身份", "requirement": "定義對象、鍵值、時間、單位、狀態、版本、關係、修正、缺失，以及跨系統同一實體如何對帳。" }, { "name": "代表性及標準答案", "requirement": "檢查使用者、期間、語言、常見與罕見案例、成果、標註、分歧、選擇偏差、洩漏及時間變化。" } ] }, { "heading": "證明數據能支援生產", "introduction": "就緒包括初次導出後持續保持證據可用的管線及控制。", "checks": [ { "name": "存取及保護", "requirement": "執行租戶、角色、對象、欄位、目的、環境及時間邊界；遮蔽或最小化敏感數據並記錄存取。" }, { "name": "管線及檢索", "requirement": "測試導入完整、解析、重複、遲到與失敗記錄、譜系、索引、檢索召回、引用、新鮮度及重處理。" }, { "name": "反饋及維護", "requirement": "收集修正、例外、成果、漂移、政策變更、責任、服務水平、事故處理及維持數據可用的成本。" } ] } ], "procedure": [ { "position": 1, "name": "定義決策", "instruction": "說明使用者、流程、輸出、記錄系統、失敗後果、新鮮度、證據及動作邊界。" }, { "position": 2, "name": "建立來源圖", "instruction": "只盤點所需欄位、語義、負責人、權限、歷史、質量、存取方法及已知缺口，而非全部企業數據。" }, { "position": 3, "name": "測試代表性樣本", "instruction": "在普通、邊界、缺失、過期、衝突及受限案例運行導入、檢索或建模，記錄未知項。" }, { "position": 4, "name": "選擇最小修復", "instruction": "只修復阻礙決策的身份、收集、標註、存取、質量規則、反饋，或改用另一流程。" } ], "frequentlyAskedQuestions": [ { "question": "開始前是否要清洗所有數據？", "answer": "不需要。由一條高價值流程開始，讓所需證據適合該決策；沒有決策邊界的全企業清洗往往浪費。" }, { "question": "歷史數據是否越多越好？", "answer": "不是。舊數據可能反映過期政策、偏差、產品、使用者或流程；相關性、覆蓋、權利、標註及已知變化更重要。" }, { "question": "合成數據可否解決缺失？", "answer": "清楚標記時可支援測試或罕見案例覆蓋，但未經驗證不能證明真實世界的發生率、行為、權利或成果。" } ] }, "zhHans": { "name": "企业怎样判断数据是否已经适合做 AI？", "summary": "数据就绪必须按照任务判断。企业不需要先把所有数据清洗到完美；它需要为一条已经定义的流程和失败后果准备正确记录、语义、权限、历史、证据标注和反馈。", "decisionSignals": [ { "name": "相关", "meaning": "连接已经定义的决策和结果" }, { "name": "可信", "meaning": "有人负责、可追溯、可解释、有权限" }, { "name": "可用", "meaning": "有代表性、可访问、可测试、可维护" } ], "sections": [ { "heading": "在一条流程中评估数据", "introduction": "数据仓库清单或行数不能证明某项 AI 决策已经就绪。", "checks": [ { "name": "来源与责任", "requirement": "说明记录系统、业务与技术负责人、收集目的、使用权限、保留、地区、导出和删除责任。" }, { "name": "语义与身份", "requirement": "定义对象、键值、时间、单位、状态、版本、关系、修正、缺失，以及跨系统同一个实体怎样对账。" }, { "name": "代表性与标准答案", "requirement": "检查用户、期间、语言、常见与罕见案例、结果、标注、分歧、选择偏差、泄漏和时间变化。" } ] }, { "heading": "证明数据能够支持生产", "introduction": "就绪包括初次导出后持续保持证据可用的管线和控制。", "checks": [ { "name": "访问与保护", "requirement": "执行租户、角色、对象、字段、目的、环境和时间边界；遮蔽或最小化敏感数据并记录访问。" }, { "name": "管线与检索", "requirement": "测试导入完整、解析、重复、迟到与失败记录、谱系、索引、检索召回、引用、新鲜度和重新处理。" }, { "name": "反馈与维护", "requirement": "收集修正、异常、结果、漂移、政策变更、责任、服务水平、事故处理和维持数据可用的成本。" } ] } ], "procedure": [ { "position": 1, "name": "定义决策", "instruction": "说明用户、流程、输出、记录系统、失败后果、新鲜度、证据和动作边界。" }, { "position": 2, "name": "建立来源图", "instruction": "只盘点所需字段、语义、负责人、权限、历史、质量、访问方法和已知缺口，而不是全部企业数据。" }, { "position": 3, "name": "测试代表性样本", "instruction": "在普通、边界、缺失、过期、冲突和受限案例运行导入、检索或建模，记录未知项。" }, { "position": 4, "name": "选择最小修复", "instruction": "只修复阻碍决策的身份、收集、标注、访问、质量规则、反馈，或改用另一流程。" } ], "frequentlyAskedQuestions": [ { "question": "开始前需要清洗所有数据吗？", "answer": "不需要。从一条高价值流程开始，让所需证据适合该决策；没有决策边界的全企业清洗经常造成浪费。" }, { "question": "历史数据越多越好吗？", "answer": "不是。旧数据可能反映过期政策、偏差、产品、用户或流程；相关性、覆盖、权利、标注和已知变化更重要。" }, { "question": "合成数据可以解决缺失吗？", "answer": "清楚标记时可以支持测试或罕见案例覆盖，但未经验证不能证明真实世界的发生率、行为、权利或结果。" } ] } }, "limitations": [ "This asset is provider-authored and is not independent validation or a claim of client results.", "It is an implementation aid, not legal, privacy, security, financial, procurement, or regulatory advice.", "Thresholds and controls must be adapted to the organisation, workflow, jurisdiction, and risk owners." ] }