{
  "schema": "insight.wwdc.priority-session.refined-zh.v1",
  "session_id": "297",
  "title": "Best practices for integrating visual intelligence in your app",
  "title_zh": "视觉智能：Best practices for integrating visual intelligence in your app",
  "priority": "P1",
  "themes": [
    "visual_intelligence_search",
    "local_ai_mlx_core_ai"
  ],
  "theme_labels_zh": [
    "视觉智能与搜索",
    "端侧 AI / MLX / Core AI"
  ],
  "positioning_zh": "Visual Intelligence 集成：把视觉理解能力嵌入 App 体验和任务链。",
  "executive_takeaway_zh": "297 体现多模态能力进入应用的关键不是识别本身，而是把识别结果转成可解释、可操作的用户价值。",
  "refined_claims": [
    {
      "id": "r1",
      "title_zh": "多模态入口",
      "claim_zh": "视觉智能让 App 能基于图像和场景上下文提供下一步行动。",
      "evidence_refs": [
        "a1",
        "a2",
        "a3"
      ],
      "evidence_policy": "safe anchor ids only; no full transcript text",
      "review_status": "editorial_refined_v1_needs_quote_check"
    },
    {
      "id": "r2",
      "title_zh": "体验约束",
      "claim_zh": "结果必须可解释、可纠错，并与用户当前任务相关，否则会变成噪声。",
      "evidence_refs": [
        "a2",
        "a3",
        "a4"
      ],
      "evidence_policy": "safe anchor ids only; no full transcript text",
      "review_status": "editorial_refined_v1_needs_quote_check"
    },
    {
      "id": "r3",
      "title_zh": "工程集成",
      "claim_zh": "需要处理图像权限、上下文范围、延迟、失败处理和 UI 呈现。",
      "evidence_refs": [
        "a3",
        "a4",
        "a5"
      ],
      "evidence_policy": "safe anchor ids only; no full transcript text",
      "review_status": "editorial_refined_v1_needs_quote_check"
    },
    {
      "id": "r4",
      "title_zh": "场景选择",
      "claim_zh": "优先选择高频、明确、可验证的视觉任务，而不是泛化识别展示。",
      "evidence_refs": [
        "a4",
        "a5",
        "a6"
      ],
      "evidence_policy": "safe anchor ids only; no full transcript text",
      "review_status": "editorial_refined_v1_needs_quote_check"
    }
  ],
  "evidence_path": {
    "brief": "data/session_briefs_zh/297.json",
    "deep_skeleton": "data/priority_session_deep_zh/297.json",
    "workbench": "workbench.html?view=sessions/297/session_297_workbench.html",
    "safe_anchors": [
      {
        "id": "a1",
        "time": 7,
        "label": "0:07",
        "sourceSegment": 0,
        "public_text_zh": "内部字幕片段 #0 · 仅公开时间锚点和安全摘要，不公开原文或完整中文翻译。",
        "apple_url": ""
      },
      {
        "id": "a2",
        "time": 207,
        "label": "3:27",
        "sourceSegment": 54,
        "public_text_zh": "内部字幕片段 #54 · 仅公开时间锚点和安全摘要，不公开原文或完整中文翻译。",
        "apple_url": ""
      },
      {
        "id": "a3",
        "time": 421,
        "label": "7:01",
        "sourceSegment": 109,
        "public_text_zh": "内部字幕片段 #109 · 仅公开时间锚点和安全摘要，不公开原文或完整中文翻译。",
        "apple_url": ""
      },
      {
        "id": "a4",
        "time": 631,
        "label": "10:31",
        "sourceSegment": 163,
        "public_text_zh": "内部字幕片段 #163 · 仅公开时间锚点和安全摘要，不公开原文或完整中文翻译。",
        "apple_url": ""
      },
      {
        "id": "a5",
        "time": 850,
        "label": "14:10",
        "sourceSegment": 218,
        "public_text_zh": "内部字幕片段 #218 · 仅公开时间锚点和安全摘要，不公开原文或完整中文翻译。",
        "apple_url": ""
      },
      {
        "id": "a6",
        "time": 1053,
        "label": "17:33",
        "sourceSegment": 272,
        "public_text_zh": "内部字幕片段 #272 · 仅公开时间锚点和安全摘要，不公开原文或完整中文翻译。",
        "apple_url": ""
      }
    ],
    "policy": "safe_editorial_summary_only_no_full_transcript_or_full_translation"
  },
  "review_questions_zh": [
    "视觉结果如何转成行动？",
    "用户如何确认或纠错？",
    "图像数据是否离开设备或进入敏感路径？"
  ],
  "publication_guidance_zh": [
    "可用于汇报页、Session 介绍和主题报告入口。",
    "发布前仍需人工核对 Apple 官方表达和少量合规短引文。",
    "不得把完整 transcript 或完整中文翻译复制到公开页。"
  ],
  "quality": {
    "level": "P0/P1 editorial refined v1",
    "requires_human_quote_review": true,
    "requires_official_term_check": true,
    "generated_at": "2026-06-22"
  }
}