The Risks tab lists every risk finding linked to the current document — both AI-detected results from automated checks and manual annotations added by your team. It is the primary place to review, refine, and sign off on what Emma (and your colleagues) flagged in the file.
For creating new findings by drawing on the PDF, see PDF viewer — Creating a manual risk. For Matrix-level risk management, see Checks and Risks in the Matrix.
Tab header and counts
The tab label shows Risks (n) where n is the number of relevant findings excluding "No risk" results. Findings marked No risk do not appear in the count or list.
If the document failed AI processing, an amber banner explains the limitation. You can still add manual findings, but automated check results may be incomplete.
Two sections
The Risks tab is divided into two areas:
Document-level recommendations
At the top, Document-Level Recommendations holds deal recommendations tied to the whole file rather than a specific check — for example "Request an updated version before closing" or "No action needed".
Each recommendation has:
Field | Description |
Type | Pre-closing, Closing action, Post-closing, Specific indemnity, Specific tax indemnity, Representations and warranties, or No action needed |
Text | The recommendation itself (markdown supported) |
To manage recommendations:
Click Add recommendation to create a new entry
Click the pencil icon to edit inline
Click Delete to remove one
Multiple recommendations can coexist — use the type field to categorise them for the Red Flag Report
Recommendations appear in the report editor and export alongside per-check findings.
Risk-level findings and recommendations
Below the document-level section, findings are grouped by risk level:
Group | What it contains |
High | Significant issues requiring attention |
Medium | Moderate concerns |
Low | Minor items worth noting |
Unspecified | Relevant content where the level needs human confirmation |
Each group is a collapsible panel showing the count of findings at that level. Expand a group to see individual findings.
Understanding a finding row
Each finding corresponds to one check (a due-diligence question Emma runs against documents). A row shows:
Element | Meaning |
Check title | The name of the check this finding belongs to |
Topic | The thematic grouping (Employment, Tax, IP, etc.) |
AI / Human indicator | Whether the latest classification came from AI or a team member |
Risk level dropdown | Change the level — updates propagate to the Matrix and report |
Summary | Description of what was found; editable inline |
Reference navigation | When the finding links to multiple locations in the PDF, step through them with previous / next |
Per-finding recommendations | Optional recommendations specific to this check result |
Edit / Delete | Open the full finding dialog or remove the finding |
Click anywhere on a finding row to:
Select it in the sidebar (highlighted state)
Scroll the PDF to the linked bounding boxes
Show the highlights for that finding on the document
When you arrive from the Matrix or Red Flag Report via a deep link, the correct group expands and the finding scrolls into view automatically.
AI vs manual findings
A summary line below the section header tells you the composition of findings in this document:
Message | Meaning |
"Only AI risks in this document that need human review" | All findings are AI-generated; none confirmed by a human yet |
"Only manual risks in this document" | Team members created all findings; AI found nothing relevant |
"AI detections and manual markings in this document" | Mix of both sources |
"No risks detected in this document yet" | Empty state — no relevant findings |
The small AI/Human icon next to this text reflects the same information visually.
Human review workflow:
Review each AI finding — read the summary and check the PDF evidence
Adjust the risk level if Emma over- or under-classified
Edit the summary to add context; saving marks the finding as human-classified
Delete false positives
Add manual findings for issues Emma missed (via the PDF viewer)
When satisfied, use Approve in the sidebar header and choose to mark AI risk levels as human-reviewed
Editing a finding
Inline summary edit — click the summary text to open the markdown editor. Saving updates the finding and marks it human-classified if it was AI-generated.
Full edit dialog — click the pencil icon to open the complete finding form where you can:
Change topic and check assignment
Adjust risk level
Edit the summary
Add or remove bounding boxes (the currently selected boxes on the PDF are included)
Create a new check on the fly if none fits
Risk level change — use the dropdown on the finding row for a quick level change without opening the dialog. The PDF highlight colour updates immediately.
Delete — remove a finding entirely. This deletes the answer for this document/check combination and removes all linked bounding boxes from the PDF.
Show and hide highlights on the PDF
Each risk-level group header has an eye icon:
Eye open — highlights for this level are visible on the PDF
Eye crossed out — highlights are hidden on the PDF (findings remain in the sidebar)
Use this to focus on High risks first without removing Medium or Low findings from the record.
Per-finding recommendations
Expand a finding to see its recommendations subsection (when present). These are check-specific recommendations — distinct from document-level recommendations at the top of the tab.
The same recommendation types apply (Pre-closing, Closing action, etc.). They flow into the Red Flag Report nested under the corresponding risk row.
Collaboration and presence
When another team member is editing a finding's summary in the report editor or the viewer simultaneously, Emma shows an Editing by [name] badge and locks the field to prevent overwrites.
Changes sync live:
Risk level changed in the viewer → updates the Matrix and open report editor
Risk edited in the report → updates the viewer if open on the same document
New manual finding → appears in the Matrix column for that check
Empty state
When no relevant findings exist, the tab shows a green shield icon with "No risks detected". This does not mean checks were not run — it means every check returned No risk or nothing relevant. Verify in the Matrix that checks are assigned to this document's type.
Bundled documents
When the open file is a bundle main document with PDF annexes, the Risks tab aggregates findings from all files in the bundle. Selecting a finding scrolls to the correct file's page within the merged PDF view.
Related workflows
Goal | Where to go |
Add a new manual finding | PDF viewer → Flag/select risks mode → draw bounding box |
Configure which checks run | |
See risks across all documents | Matrix → Risks tab |
Write up findings for export | Red Flag Report editor |
Turn a key term into a risk | Key terms tab → Identify as risk |
If you need additional help with the Risks tab or encounter any other problem, don't hesitate to reach out through the chat bubble in the bottom right or contact [email protected].