Decision owner
HSE manager
The named role remains accountable for the consequential decision.
Oil, Gas & Mining
Designed for HSE specialist
Turn safety incidents, near misses, and observations into transferable learning by clustering causal patterns across sites and linking them to the standards and permits that should have prevented them, with HSE approval before actions are issued.
Contract OIL-02 · primary aggregate SafetyLearningCase · mode loading
Decision brief
This brief starts with the decision, evidence, and accountable owner. The guided workflow then shows how the recommendation was produced.
The evidence and control points that shape this recommendation.
Decision owner
HSE manager
The named role remains accountable for the consequential decision.
Governed evidence
6 record types
Structured records and cited text ground the recommendation.
Workflow
5 controlled stages
Each transition records its input, output, status, and audit context.
Human controls
1 approval gate
Approval gates prevent the copilot from making consequential decisions.
A dropped-object near miss at a mine conveyor is classified as high potential; the assistant clusters it with 11 similar events across three sites where the same lifting permit omitted an exclusion-zone control, maps the gap to the lifting standard clause, and the HSE manager approves a permit template change with verification at all affected sites..
Turn safety incidents, near misses, and observations into transferable learning by clustering causal patterns across sites and linking them to the standards and permits that should have prevented them, with HSE approval before actions are issued.
Issued actions with verification plans gives HSE manager the evidence needed to begin the final governed business outcome, with the decision trail retained for review.
Step 1 of 5
| Field | Value |
|---|
| Field | Value |
|---|
The output appears here immediately. More technical evidence remains available below.
The demonstration keeps deterministic controls, AI assistance, and approval separate.
Source systems enforce identity, workflow order, policy filters, and record updates.
Retrieval and generation assemble relevant evidence into a cited recommendation or draft.
HSE manager verifies material claims and approves the consequential action.
Run the stages in order before using the Copilot. Every stage explains what happens, why it matters, the evidence it uses, and the complete API request and response.
Automated stage
What happens: Field worker or supervisor performs capture the incident or observation using Incident report, near miss, or hazard observation and produces Structured record with activity and energy source. The transition is deterministic and does not require autonomous model approval.
Why it matters: Structured record with activity and energy source gives Classification agent the evidence needed to begin Classify severity and control failure, with the decision trail retained for review.
Loading relevant sample records.
Ready to run.
POST /api/domain/workflow/capture-the-incident-or-observation
Run this API to see its formatted outcome.
| Field | Value |
|---|
| Field | Value |
|---|
Automated stage
What happens: Classification agent performs classify severity and control failure using Event narrative and permit record and produces Actual and potential severity with failed control category. The transition is deterministic and does not require autonomous model approval.
Why it matters: Actual and potential severity with failed control category gives Pattern analysis agent the evidence needed to begin Cluster causal patterns across sites, with the decision trail retained for review.
Loading relevant sample records.
Ready to run.
POST /api/domain/workflow/classify-severity-and-control-failure
Run this API to see its formatted outcome.
| Field | Value |
|---|
| Field | Value |
|---|
Automated stage
What happens: Pattern analysis agent performs cluster causal patterns across sites using Historic incidents and observations and produces Causal themes and recurrence evidence. The transition is deterministic and does not require autonomous model approval.
Why it matters: Causal themes and recurrence evidence gives Standards mapping agent the evidence needed to begin Map themes to standards and controls, with the decision trail retained for review.
Loading relevant sample records.
Ready to run.
POST /api/domain/workflow/cluster-causal-patterns-across-sites
Run this API to see its formatted outcome.
| Field | Value |
|---|
| Field | Value |
|---|
Automated stage
What happens: Standards mapping agent performs map themes to standards and controls using Causal themes and safety standard clauses and produces Gap analysis between expected and actual controls. The transition is deterministic and does not require autonomous model approval.
Why it matters: Gap analysis between expected and actual controls gives HSE manager the evidence needed to begin Approve and issue learning actions, with the decision trail retained for review.
Loading relevant sample records.
Ready to run.
POST /api/domain/workflow/map-themes-to-standards-and-controls
Run this API to see its formatted outcome.
| Field | Value |
|---|
| Field | Value |
|---|
Human approval required
What happens: HSE manager performs approve and issue learning actions using Proposed actions with target sites and produces Issued actions with verification plans. A named human approver must review the evidence before the transition completes.
Why it matters: Issued actions with verification plans gives HSE manager the evidence needed to begin the final governed business outcome, with the decision trail retained for review.
Loading relevant sample records.
Ready to run.
POST /api/domain/workflow/approve-and-issue-learning-actions
Run this API to see its formatted outcome.
| Field | Value |
|---|
| Field | Value |
|---|
Run a stage to see its explanation, evidence, full request, and full response.
Ask a question and receive an answer grounded in retrieved SafetyLearningCase evidence.
Select a starter question or enter your own.
Loaded live from /api/config, /api/domain/model, and
/api/domain/sample.
safetyLearningCases/tenantId + /operatingSiteIdA dropped-object near miss at a mine conveyor is classified as high potential; the assistant clusters it with 11 similar events across three sites where the same lifting permit omitted an exclusion-zone control, maps the gap to the lifting standard clause, and the HSE manager approves a permit template change with verification at all affected sites.
| Container | Role | Hierarchical partition key | Document types |
|---|---|---|---|
safetyLearningCases |
Primary | /tenantId + /operatingSiteId |
SafetyLearningCase, IncidentReport, SafetyObservation, PermitRecord, LearningAction |
safetyStandardsLibrary |
Reference | /tenantId + /standardId |
SafetyStandardClause |
| Type | Container | Partition key | Identifier pattern | Embedding fields |
|---|---|---|---|---|
SafetyLearningCase |
safetyLearningCases |
/tenantId + /operatingSiteId |
learningcase::{operatingSiteId}::{caseId} |
None |
IncidentReport |
safetyLearningCases |
/tenantId + /operatingSiteId |
incident::{operatingSiteId}::{incidentId} |
eventNarrativeVector |
SafetyObservation |
safetyLearningCases |
/tenantId + /operatingSiteId |
observation::{operatingSiteId}::{observationId} |
observationVector |
PermitRecord |
safetyLearningCases |
/tenantId + /operatingSiteId |
permit::{operatingSiteId}::{permitId} |
None |
SafetyStandardClause |
safetyStandardsLibrary |
/tenantId + /standardId |
standard::{standardId}::{clauseRef} |
clauseVector |
LearningAction |
safetyLearningCases |
/tenantId + /operatingSiteId |
action::{operatingSiteId}::{actionId} |
actionVector |
Read the case with its incidents, observations, and permits in one site partition query.
safetyLearningCases
single-partition
SELECT * FROM c WHERE c.tenantId = @tenantId AND c.operatingSiteId = @operatingSiteId AND c.type IN ('SafetyLearningCase','IncidentReport','SafetyObservation','PermitRecord') ORDER BY c.occurredUtc DESC
Partition-key fan-out control-plane query: surface low-consequence events whose potential severity was severe. It cannot carry the second partition key because learning value comes precisely from comparing near misses across sites, so it runs as a tenant-scoped scheduled sweep off the interactive path, bounded by near-miss classifications inside a 90-day occurredUtc window, executed with MaxItemCount paging and continuation tokens.
safetyLearningCases
partition-key-fan-out
SELECT * FROM c WHERE c.tenantId = @tenantId AND c.type = 'IncidentReport' AND c.outcomeClassification = 'nearMiss' AND c.occurredUtc > @lookbackUtc ORDER BY c.occurredUtc DESC
Count observations by hazard category for a specific high-risk activity type.
safetyLearningCases
single-partition
SELECT * FROM c WHERE c.tenantId = @tenantId AND c.operatingSiteId = @operatingSiteId AND c.type = 'SafetyObservation' AND c.hazardCategory = @hazardCategory AND c.reportedUtc > @lookbackUtc ORDER BY c.reportedUtc DESC
Partition-key fan-out control-plane query: list issued learning actions whose verification is still outstanding. It cannot carry the second partition key because the HSE assurance report is a multi-site compliance view, so it runs as a tenant-scoped scheduled sweep off the interactive path, bounded by issued actions with no verifiedUtc inside the 60-day closure target, executed with MaxItemCount paging and continuation tokens.
safetyLearningCases
partition-key-fan-out
SELECT * FROM c WHERE c.tenantId = @tenantId AND c.type = 'LearningAction' AND c.approvalState = 'issued' AND c.verificationState = 'outstanding' ORDER BY c.actionId ASC