{
  "schemaVersion": "1.0",
  "title": "AI Observability and Evaluation Operations System",
  "lang": "en",
  "prd": {
    "oneLiner": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
    "goal": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
    "whyNow": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
    "category": "Internal operations system",
    "platforms": [
      "Web"
    ],
    "problem": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
    "solution": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
    "alternatives": "Email, chat, spreadsheets, and manual ledgers",
    "differentiator": "Manage intake, processing, and operations data as one connected source of truth.",
    "targets": [
      {
        "name": "Product, ML, and operations practitioners managing AI quality, cost, and safety",
        "role": "Requester and operational user",
        "needs": "Quickly register needed work and know its status.",
        "pain": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change."
      },
      {
        "name": "AI platform and ML operations coordinator",
        "role": "Operations administrator",
        "needs": "Manage priority and history and review operations metrics.",
        "pain": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change."
      },
      {
        "name": "System and policy administrator",
        "role": "Operations and security administrator",
        "needs": "Manage access, policy, and audit history.",
        "pain": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change."
      }
    ],
    "scenarios": [
      {
        "text": "An operational user registers required work and checks its status.",
        "start": "P1"
      },
      {
        "text": "An operator processes queued work and handles exceptions.",
        "start": "P5"
      },
      {
        "text": "System and policy administrator reviews policy and access.",
        "start": "P9"
      },
      {
        "text": "System and policy administrator reviews audit records and operating metrics.",
        "start": "P10"
      }
    ],
    "northStar": "Share of operations requests completed on time",
    "kpis": [
      {
        "name": "On-time processing rate",
        "target": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
        "baseline": "Not measured",
        "method": "Monthly aggregation of state changes against target dates",
        "refs": [
          "F6"
        ]
      },
      {
        "name": "Operations visibility",
        "target": "100% of key work tracked",
        "baseline": "Not measured",
        "method": "Share of records connected to reporting",
        "refs": [
          "F11"
        ]
      },
      {
        "name": "Auditable operations rate",
        "target": "100% of key changes retained",
        "baseline": "Not measured",
        "method": "Monthly audit-log and change-history review",
        "refs": [
          "F12"
        ]
      }
    ],
    "inScope": [
      "Trace, prompt, model version, evaluation set, and cost-baseline registration",
      "Execution trace, prompt, model version, evaluation result, token, and cost-history lookup",
      "Evaluation-criteria setup, execution observation, quality, safety, and cost review, model-or-prompt change approval, and re-evaluation handling",
      "Response quality, evaluation pass rate, safety violation, latency, token usage, and model-cost reporting",
      "Trace, prompt, model version, evaluation set, and cost-baseline registration validation and duplicate prevention",
      "Search, filter, and history management",
      "Owner, priority, and SLA management",
      "Approval and policy review",
      "Exception, rejection, and reprocessing",
      "Notifications and subscriptions",
      "External integration and data synchronization",
      "Role, permission, and audit management"
    ],
    "nonGoals": [
      "External customer portal",
      "Replacing adjacent systems such as accounting or payroll"
    ],
    "assumptions": [
      "User and organization data are available through company SSO.",
      "Operators manage policy and classification criteria."
    ],
    "risks": [
      "Weak initial classification can cause incorrect processing.",
      "Stale source data can reduce operational trust."
    ],
    "openQuestions": [
      "What defines urgent processing and approval?",
      "What are the post-completion verification and reopen rules?"
    ],
    "constraints": [
      "State, owner, and priority changes require audit logs.",
      "Access is limited by role.",
      "Permission, policy, and state changes require audit logs."
    ]
  },
  "requirements": [
    {
      "id": "R1",
      "title": "Intake and data quality",
      "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
      "status": "todo",
      "priority": "high",
      "acceptance": [
        {
          "text": "Required context, changes, ownership, and status are retained in an auditable record.",
          "done": false
        }
      ],
      "features": [
        {
          "id": "F1",
          "title": "Trace, prompt, model version, evaluation set, and cost-baseline registration",
          "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Trace, prompt, model version, evaluation set, and cost-baseline registration core processing rule",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Trace, prompt, model version, evaluation set, and cost-baseline registration exception and audit handling",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        },
        {
          "id": "F2",
          "title": "Trace, prompt, model version, evaluation set, and cost-baseline registration validation and duplicate prevention",
          "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Trace, prompt, model version, evaluation set, and cost-baseline registration validation and duplicate prevention core processing rule",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Trace, prompt, model version, evaluation set, and cost-baseline registration validation and duplicate prevention exception and audit handling",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": "R2",
      "title": "Records and search",
      "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
      "status": "todo",
      "priority": "high",
      "acceptance": [
        {
          "text": "Required context, changes, ownership, and status are retained in an auditable record.",
          "done": false
        }
      ],
      "features": [
        {
          "id": "F3",
          "title": "Execution trace, prompt, model version, evaluation result, token, and cost-history lookup",
          "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Execution trace, prompt, model version, evaluation result, token, and cost-history lookup core processing rule",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Execution trace, prompt, model version, evaluation result, token, and cost-history lookup exception and audit handling",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        },
        {
          "id": "F4",
          "title": "Search, filter, and history management",
          "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Search, filter, and history management core processing rule",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Search, filter, and history management exception and audit handling",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": "R3",
      "title": "Processing and ownership",
      "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
      "status": "todo",
      "priority": "high",
      "acceptance": [
        {
          "text": "Required context, changes, ownership, and status are retained in an auditable record.",
          "done": false
        }
      ],
      "features": [
        {
          "id": "F5",
          "title": "Evaluation-criteria setup, execution observation, quality, safety, and cost review, model-or-prompt change approval, and re-evaluation handling",
          "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Evaluation-criteria setup, execution observation, quality, safety, and cost review, model-or-prompt change approval, and re-evaluation handling core processing rule",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Evaluation-criteria setup, execution observation, quality, safety, and cost review, model-or-prompt change approval, and re-evaluation handling exception and audit handling",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        },
        {
          "id": "F6",
          "title": "Owner, priority, and SLA management",
          "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Owner, priority, and SLA management core processing rule",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Owner, priority, and SLA management exception and audit handling",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": "R4",
      "title": "Approval and exceptions",
      "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
      "status": "todo",
      "priority": "high",
      "acceptance": [
        {
          "text": "Required context, changes, ownership, and status are retained in an auditable record.",
          "done": false
        }
      ],
      "features": [
        {
          "id": "F7",
          "title": "Approval and policy review",
          "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
          "status": "todo",
          "priority": "mid",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Approval and policy review core processing rule",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Approval and policy review exception and audit handling",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        },
        {
          "id": "F8",
          "title": "Exception, rejection, and reprocessing",
          "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
          "status": "todo",
          "priority": "mid",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Exception, rejection, and reprocessing core processing rule",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Exception, rejection, and reprocessing exception and audit handling",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": "R5",
      "title": "Notifications and integrations",
      "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
      "status": "todo",
      "priority": "mid",
      "acceptance": [
        {
          "text": "Required context, changes, ownership, and status are retained in an auditable record.",
          "done": false
        }
      ],
      "features": [
        {
          "id": "F9",
          "title": "Notifications and subscriptions",
          "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Notifications and subscriptions core processing rule",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Notifications and subscriptions exception and audit handling",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        },
        {
          "id": "F10",
          "title": "External integration and data synchronization",
          "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "External integration and data synchronization core processing rule",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "External integration and data synchronization exception and audit handling",
              "desc": "Connect traces, prompts, model versions, evaluations, costs, review evidence, and change-approval and re-evaluation results to support AI operating decisions. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": "R6",
      "title": "Insights and controls",
      "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
      "status": "todo",
      "priority": "mid",
      "acceptance": [
        {
          "text": "Required context, changes, ownership, and status are retained in an auditable record.",
          "done": false
        }
      ],
      "features": [
        {
          "id": "F11",
          "title": "Response quality, evaluation pass rate, safety violation, latency, token usage, and model-cost reporting",
          "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
          "status": "todo",
          "priority": "high",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Response quality, evaluation pass rate, safety violation, latency, token usage, and model-cost reporting core processing rule",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Response quality, evaluation pass rate, safety violation, latency, token usage, and model-cost reporting exception and audit handling",
              "desc": "Prompts, model versions, execution traces, evaluation results, and cost data are disconnected, making it hard to judge whether an AI quality or cost change came from a specific change. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        },
        {
          "id": "F12",
          "title": "Role, permission, and audit management",
          "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
          "status": "todo",
          "priority": "mid",
          "acceptance": [
            {
              "text": "Required context, changes, ownership, and status are retained in an auditable record.",
              "done": false
            }
          ],
          "specs": [
            {
              "title": "Role, permission, and audit management core processing rule",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            },
            {
              "title": "Role, permission, and audit management exception and audit handling",
              "desc": "Keep pre-release evaluation pass rates at 95% or higher and identify quality regression or cost spikes before deployment. Retain change, failure, and reprocessing conditions with history.",
              "acceptance": [
                {
                  "text": "Required context, changes, ownership, and status are retained in an auditable record.",
                  "done": false
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "ia": {
    "sections": [
      {
        "id": "S1",
        "title": "Start and intake",
        "pages": [
          {
            "id": "P1",
            "title": "Operations home",
            "type": "top",
            "surface": "screen",
            "refs": [],
            "children": [
              {
                "id": "P2",
                "title": "Work inbox",
                "type": "page",
                "surface": "screen",
                "refs": [
                  "F1"
                ],
                "children": [
                  {
                    "id": "P3",
                    "title": "Create work item",
                    "type": "page",
                    "surface": "modal",
                    "refs": [
                      "F1:0",
                      "F1:1"
                    ],
                    "children": []
                  },
                  {
                    "id": "P11",
                    "title": "Input review",
                    "type": "page",
                    "surface": "panel",
                    "refs": [
                      "F2",
                      "F2:0",
                      "F2:1"
                    ],
                    "children": []
                  }
                ]
              }
            ]
          }
        ]
      },
      {
        "id": "S2",
        "title": "Records and work execution",
        "pages": [
          {
            "id": "P4",
            "title": "Record detail",
            "type": "top",
            "surface": "screen",
            "refs": [
              "F3",
              "F3:0"
            ],
            "children": [
              {
                "id": "P12",
                "title": "Activity timeline",
                "type": "page",
                "surface": "panel",
                "refs": [
                  "F3:1"
                ],
                "children": []
              },
              {
                "id": "P13",
                "title": "Search and filters",
                "type": "page",
                "surface": "drawer",
                "refs": [
                  "F4",
                  "F4:0",
                  "F4:1"
                ],
                "children": []
              }
            ]
          },
          {
            "id": "P5",
            "title": "Processing workspace",
            "type": "page",
            "surface": "screen",
            "refs": [
              "F5",
              "F5:0"
            ],
            "children": [
              {
                "id": "P14",
                "title": "Processing exception",
                "type": "page",
                "surface": "drawer",
                "refs": [
                  "F5:1"
                ],
                "children": []
              },
              {
                "id": "P15",
                "title": "Owner and due date",
                "type": "page",
                "surface": "panel",
                "refs": [
                  "F6",
                  "F6:0",
                  "F6:1"
                ],
                "children": []
              }
            ]
          }
        ]
      },
      {
        "id": "S3",
        "title": "Review and recovery",
        "pages": [
          {
            "id": "P6",
            "title": "Review queue",
            "type": "top",
            "surface": "screen",
            "refs": [
              "F7",
              "F7:0"
            ],
            "children": [
              {
                "id": "P16",
                "title": "Decision detail",
                "type": "page",
                "surface": "panel",
                "refs": [
                  "F7:1"
                ],
                "children": []
              },
              {
                "id": "P17",
                "title": "Reject and reprocess",
                "type": "page",
                "surface": "modal",
                "refs": [
                  "F8",
                  "F8:0",
                  "F8:1"
                ],
                "children": []
              }
            ]
          }
        ]
      },
      {
        "id": "S4",
        "title": "Notifications and integrations",
        "pages": [
          {
            "id": "P7",
            "title": "Notification center",
            "type": "top",
            "surface": "screen",
            "refs": [
              "F9",
              "F9:0"
            ],
            "children": [
              {
                "id": "P18",
                "title": "Subscription settings",
                "type": "page",
                "surface": "panel",
                "refs": [
                  "F9:1"
                ],
                "children": []
              }
            ]
          },
          {
            "id": "P19",
            "title": "External integrations",
            "type": "page",
            "surface": "screen",
            "refs": [
              "F10",
              "F10:0"
            ],
            "children": [
              {
                "id": "P20",
                "title": "Synchronization failures",
                "type": "page",
                "surface": "drawer",
                "refs": [
                  "F10:1"
                ],
                "children": []
              }
            ]
          }
        ]
      },
      {
        "id": "S5",
        "title": "Insights and administration",
        "pages": [
          {
            "id": "P8",
            "title": "Operations reports",
            "type": "top",
            "surface": "screen",
            "refs": [
              "F11",
              "F11:0"
            ],
            "children": [
              {
                "id": "P21",
                "title": "Report filters",
                "type": "page",
                "surface": "drawer",
                "refs": [
                  "F11:1"
                ],
                "children": []
              }
            ]
          },
          {
            "id": "P9",
            "title": "Policy and access",
            "type": "page",
            "surface": "screen",
            "refs": [
              "F12",
              "F12:0"
            ],
            "children": [
              {
                "id": "P10",
                "title": "Audit history",
                "type": "page",
                "surface": "panel",
                "refs": [
                  "F12:1"
                ],
                "children": []
              }
            ]
          }
        ]
      }
    ]
  },
  "flow": {
    "start": "P1",
    "transitions": [
      {
        "from": "P1",
        "to": "P2",
        "label": "View work"
      },
      {
        "from": "P2",
        "to": "P3",
        "ref": "F1"
      },
      {
        "from": "P3",
        "to": "P11",
        "ref": "F2"
      },
      {
        "from": "P11",
        "to": "P4",
        "label": "Open after review"
      },
      {
        "from": "P4",
        "to": "P12",
        "ref": "F3"
      },
      {
        "from": "P4",
        "to": "P13",
        "ref": "F4"
      },
      {
        "from": "P13",
        "to": "P4",
        "label": "Select result"
      },
      {
        "from": "P4",
        "to": "P5",
        "ref": "F5"
      },
      {
        "from": "P5",
        "to": "P14",
        "ref": "F5:1"
      },
      {
        "from": "P5",
        "to": "P15",
        "ref": "F6"
      },
      {
        "from": "P15",
        "to": "P6",
        "label": "Request review"
      },
      {
        "from": "P6",
        "to": "P16",
        "ref": "F7"
      },
      {
        "from": "P16",
        "to": "P17",
        "ref": "F8"
      },
      {
        "from": "P17",
        "to": "P5",
        "label": "Reprocess"
      },
      {
        "from": "P6",
        "to": "P7",
        "ref": "F9"
      },
      {
        "from": "P7",
        "to": "P18",
        "ref": "F9:1"
      },
      {
        "from": "P7",
        "to": "P19",
        "ref": "F10"
      },
      {
        "from": "P19",
        "to": "P20",
        "ref": "F10:1"
      },
      {
        "from": "P5",
        "to": "P8",
        "ref": "F11"
      },
      {
        "from": "P8",
        "to": "P21",
        "ref": "F11:1"
      },
      {
        "from": "P1",
        "to": "P9",
        "ref": "F12"
      },
      {
        "from": "P9",
        "to": "P10",
        "ref": "F12:1"
      },
      {
        "from": "P8",
        "to": "P1",
        "label": "Refresh operations status"
      }
    ]
  }
}