Start with the right workspace
What this release adds
Investigate Pro now brings together an interactive cause canvas, linked evidence and action inspection, solution comparison, investigation completeness checks and recurring-cause analysis. These tools support human investigation and independent review. They do not establish causation automatically.
1. Confirm access
Ask your Sentinel administrator to allocate Investigate and Investigate Pro under Company products. Sign in to the company portal and choose the correct company and assessment. An active licence and editing role are needed to change records. Read-only users can inspect and export existing analysis.
2. Open Investigate Pro
Select Investigate, then Investigate Pro. Choose an investigation. If it has no Pro brief, select Investigation brief and record the objective, scope, lead, team, expected and actual conditions, review arrangements and limitations. Existing 5 Whys findings remain in the same investigation.
3. Practise with Northbank
In an authorised practice workspace, choose Load fictional Pro example. Confirm any unsaved changes before continuing. This creates a new draft containing a near miss, competing explanations, evidence references, action plans and two solution options. It does not approve an investigation or contact external services.
Know the limits
The public website demonstration is read-only. The company workbench supports editing. The existing full suite guide and Pro handbook remain useful; this companion covers the additions in version 1.3.0.
Arrange and inspect the cause map
1. Read the status first
Supported findings, hypotheses, refuted explanations and open questions have distinct labels and colours. Each cause shows its number of supporting sources. A label records the investigator's assessment; it is not a calculated probability.
2. Arrange the map
Drag a cause box to a useful position. Drag the background or use scrollbars to pan. Use Zoom in, Zoom out, Fit map or 100% to change the view. All findings in the investigation are rendered, up to the existing 500-record collection limit. Large maps may be easier to inspect at a reduced zoom.
3. Use the keyboard
Press Tab until a cause receives focus. Press Enter or Space to inspect it. Arrow keys move an editable cause by ten units; hold Shift for fifty units. Read-only users cannot change positions. Use Undo move to reverse a recent layout change, or Auto-arrange to restore the default grid. The grid is not a causal ranking.
4. Save the layout
Select Save changes. Positions are saved in the workspace and use its normal revision and collaboration checks. Undo is temporary, limited to recent moves and cleared when switching workspace or investigation. Changing layout does not change evidence or finding status.
5. Connect causes
Select Add cause to record a finding, or Connect causes to choose the contributing and resulting findings. Explain the relationship and select supporting evidence. Self-links, cycles and links to another investigation are rejected. Arrow direction is cause to resulting finding. Position alone never creates a causal connection.
Open evidence and actions from a cause
1. Select a cause
Click a cause, or press Enter while it has keyboard focus. The detail window shows its status, category and rationale. It gathers the finding's evidence plus supporting and contradicting evidence referenced by its hypothesis tests.
2. Inspect sources
Read the source reference, reliability assessment, summary and limitations. Source references point to your controlled records; this feature does not upload evidence attachments or grant access to a separate document repository.
3. Review competing tests
Read the test result and interpretation. A supporting test does not automatically promote the finding to Supported. A refuting test against a supported finding is highlighted by the completeness and readiness checks. Resolve the conflict in the underlying records.
4. Follow corrective actions
The same window lists actions linked to this finding, with owners, status and due dates. Recorded effectiveness reviews appear beneath their action. No review is shown as no review, never as evidence of success.
5. Edit with permission
Use Open / edit to open the underlying record. View-only users receive a read-only record form. Apply changes and save the workspace. Reopen the cause to inspect the revised information.
Compare solutions before committing
1. Add alternatives
In Compare solution options, select Solution option. Name the option and choose the finding it addresses. Record several plausible options when a comparison would improve the decision; retaining a rejected option helps explain the final choice.
2. Record estimates
Choose estimated effectiveness, estimated cost, implementation effort and timeframe. Use Unassessed or Unknown when the basis is missing. These are qualitative judgements, not a financial calculation or validated forecast. Record the assessment basis and uncertainties in Assumptions.
3. Explain the decision
Choose Candidate, Selected or Rejected and write the rationale. Avoid selecting a weak option just because it is cheap. Compare how each option addresses the supported cause, its practical constraints and any new risks.
4. Link the action
A selected option needs a corrective action linked to the same finding. Create that action in Actions, assign an owner and due date, then select it in the solution form. The server rejects an action linked to another finding. Add an action assurance plan with success criteria, baseline and review date.
5. Verify later
The comparison table keeps Estimated effectiveness separate from Actual reviews. Record observed outcomes through Assure effectiveness reviews. A completed action, selected option or High estimate is not proof that the change worked.
Check completeness and recurring causes
1. Inspect completeness
The completeness card counts supported findings with supporting tests, supported findings with refuting tests, findings without supporting sources, actions lacking an owner or due date, and outstanding readiness requirements. Open the listed findings to inspect the gaps.
2. Complete the review requirements
The separate readiness checklist explains the requirements before submission. Selected solution assessments must have an action, estimates, assumptions and rationale. These requirements are checked on the server as well as in the interface. A complete checklist is not approval.
3. Refresh authorised investigations
Select Refresh authorised investigations. The tool reads saved workspaces you may access in the current company. It uses the current workspace's unsaved working copy where applicable. Practice mode reads only the local practice records. Nothing is sent to an AI service.
4. Interpret the results
Supported findings are grouped by their recorded contributing-condition category. The table shows distinct investigations and the underlying finding references. A repeated category suggests where to look; it does not prove a shared root cause. Refuted and untested findings are excluded. A second table identifies supported descriptions repeated across investigations, ignoring case and punctuation. It does not infer semantic equivalence and may miss differently worded findings.
5. Check coverage and freshness
Read the workspace count, excluded or unavailable count and snapshot time. Results are limited to your access, not necessarily the whole company. Refresh after changes. Switching workspace or company clears the previous result; in-flight results from the previous context are discarded.
Run the Northbank exercise and share findings
1. Explore the map
Open the Northbank practice draft. Inspect the supported route-crossing finding, the process omission, the refuted persistent-brake hypothesis and the open delivery-pressure question. Move the process omission near the crossing finding and save. Export the causal map as SVG to confirm your arrangement is retained.
2. Compare the options
Inspect Restore and inspect physical separation and Brief workers without changing the route. The first is selected and linked to an action; the second is rejected with a reason. These are fictional estimates. Neither option claims a measured outcome.
3. Deliberately leave a gap
In a disposable practice draft, change the selected option's cost to Unknown. The record can remain a draft, but the readiness checklist will require the selected assessment to be completed before review. Restore the assessment after inspecting the warning.
4. Review the report
Use Export, then Report. The HTML report includes the causal map, solution records, evidence and other investigation records. Open it in your browser and select Print / Save as PDF. The map SVG is also available separately. Recurring-cause snapshots are a live workbench view and are not included in the single-workspace report.
5. Carry the learning forward
Save and submit for a separate authorised reviewer. Supported findings can be proposed to Bowtie through the existing reviewed transfer. Use Assure to inspect the resulting controls and review actual action effectiveness. Keep uncertainties and limitations in every stage.
Administration and rollout
Install version 1.3.0 as an update to the existing plugin. Existing schema-two records gain empty layout and solution collections without losing their identifiers. Company allocation and annual seats remain unchanged. Test the workflow with your investigation team before wider operational use.