flowchart TD C["Campaigns / Triage"] --> F["Findings"] F --> I["Investigate"] I --> A["Actions"] A --> H["Campaign History"] C -. "Related view" .-> E["Evidence Explorer"] A1["Kav AI Assistant"] -. "Visible campaign and<br/>finding context" .-> C A1 -.-> F A1 -.-> I A1 -.-> A
18 Integrity Engineer workflow
Triage campaign evidence into governed action
Integrity Engineer is where an engineer reviews campaign evidence, prioritizes and investigates findings, creates KAP work records, and follows the campaign’s history. The engineer remains responsible for the integrity decision; the workspace does not turn evidence or Assistant output into an automatic action.
18.1 Typical workflow
The workspace is a network of related views, not a forced wizard. Evidence Explorer is available when the engineer needs map, gallery, or 3D detail, then returns the user to the triage context. The Assistant can help from any view without becoming a workflow stage of its own.
18.2 Workspace surfaces
| Surface | User question | Role in the workflow |
|---|---|---|
| Campaigns / Triage | What in this campaign needs attention now? | Start with a compact campaign triage queue. Open the evidence explorer only when detailed spatial or image review is needed. |
| Findings | Which candidates should I review first? | Use the campaign-scoped, severity-ordered queue to select a finding. |
| Investigate | What evidence supports or contradicts this finding? | Examine selected evidence and carry the chosen candidate explicitly into the next action. |
| Actions | What governed KAP work should be recorded? | Create and manage KAP work orders, with accurately labelled Markdown and CSV exports. External IDMS handoff remains unavailable until a connector exists. |
| Campaign History | What has happened in this campaign? | Review campaign chronology. It is not yet a complete immutable decision audit trail. |
| Kav AI Assistant | What can I learn or navigate to in this context? | Available throughout the workspace with visible organization, campaign, and optional finding grounding. |
18.3 Review a campaign
18.3.1 Establish campaign scope
Enter a valid organization and campaign scope. Campaign-required views resolve an available campaign before presenting triage, so the engineer begins with a meaningful review boundary rather than an empty surface. Changing campaign clears any finding that no longer belongs to it.
18.3.2 Triage and select a finding
Begin in Campaigns / Triage with the highest-value campaign information and a compact review queue. Move to Findings to prioritize candidates. When detailed context is needed, use the related Evidence Explorer for map, gallery, or 3D inspection, then return to triage or a selected finding.
18.3.3 Investigate and create action
Use Investigate to validate a selected finding against its supporting evidence. Continue to Actions to create a reviewable KAP work order. The user-visible export describes its actual format, and the IDMS control stays disabled until a real connector is available; KAP must not imply a completed external handoff when none occurred.
18.3.4 Review campaign history
Use Campaign History to understand the current campaign chronology. A complete, immutable record of finding decisions, exports, and external handoffs is a target capability and must not be claimed before those persisted events exist.
18.3.5 Ask in context
The Kav AI Assistant is one shared, scope-aware panel, available from workspace chrome, the mobile launcher, the sidebar, and compatible /ask links. It shows its grounding context and may propose reviewable view actions, but it cannot silently change scope, make a lifecycle decision, export data, or contact an external system.
18.4 Current and target decision boundary
| State | Engineer responsibility |
|---|---|
| Current | Review campaign evidence, choose a finding, investigate it, and create or export a KAP-local work order. Campaign History provides chronology. |
| Target | Persist Confirm, Dismiss, or Reclassify decisions with rationale; maintain an immutable audit trail; and hand off approved work through a connected external system with explicit confirmation. |
Q3 sensor-native analysis proposes independent RGB or thermal image candidates for human review. Automated cross-modal detection, evidence-confidence gates, and quantified inspection prioritization are Q4 targets. None of those targets authorizes automatic equipment control or an unreviewed integrity decision.
18.5 Relationship to Data Explorer
Campaign evidence is prepared in Data Explorer. Today, campaign presence is the implicit handoff. The planned evidence-package readiness transition will let an engineer see what was submitted, acknowledge it, request clarification, and preserve the review history.