A presentation is a curated collection of findings drawn from across sessions and diagnostics. Presentations let users assemble findings from many separate investigations into a coherent narrative, and keep a link back to the session each finding came from.
Creating a Presentation
Presentations can be created manually or automatically by a scan. When a scan finishes all of its diagnostics, it automatically builds a presentation by snapshotting findings from every completed session and using the AI to organize them into slides. Scan-generated presentations are scoped to the scan's entity and vintage.
The Presentations list (accessible via the sidebar) shows all presentations you own, can edit, or can view, sorted by last updated. It is sortable by any column and filterable by name, description, entity, vintage, and owner.
To create a presentation manually, click New Presentation. Provide:
- Name — A descriptive title for the presentation.
- Description — Optional context about what the presentation covers.
- Entity — Optional entity to associate with the presentation, indicating which organization or partner the findings relate to.
- Vintage — Optional free-form text identifying the data period (e.g., "2024 Q3", "CY2025").
- Tags — Optional tags for categorization. Users can select from existing tags or create new ones. Tags are displayed as pills on the presentation list and detail views.
- Editors — Compass Author-level (or higher) users or user groups granted full edit access to the presentation, with the same rights as the owner.
- Viewers — Users or user groups granted read-only access to the presentation.
Adding Findings
Findings can be added to presentations in two ways:
- From the finding menu — Each finding has an Add button that opens a dropdown listing all presentations you own or can edit. Select a presentation to add the finding to it. You can also create a new presentation directly from this dropdown by clicking New Presentation, typing a name, and pressing Enter — the presentation is created and the finding is added to it in one step, without leaving the session.
- From the finding edit panel — When editing a finding, a Presentations section lists all presentations you own or can edit. Check a presentation to add the finding; uncheck to remove it.
A single finding can belong to multiple presentations.
Adding a finding to a presentation copies it — the finding's title, description, query, results, and chart configuration as they stand at that moment. The presentation's copy is independent from then on: editing the finding in its session does not change the presentation, and restyling the finding in the builder does not change the session. To pull in newer data, use Refresh Data in Presentation.
Viewing a Presentation
The presentation detail page displays findings organized into slides, stepped through one at a time. Each slide renders its findings according to the slide's layout (see Slide Layouts below). If no layout has been set, the slide falls back to a simple list of result titles.
Each finding's title includes a link icon that opens the source session in a new tab, allowing viewers to drill down into the context behind a finding. If the viewer does not have access to the source session's diagnostic, the icon is still visible but non-clickable, with a tooltip indicating who to contact for access.
Sharing
Presentations are private by default — only the owner can see them. To share a presentation, add individual users or user groups as editors or viewers on the presentation form.
- Editors have the same rights as the owner: they can edit the presentation, use the builder, manage findings, and manage who else has access. Only Compass Author-level users (or higher) can be added as individual editors; any user group can be added as an editor group. When a user group is added as an editor, all members of that group receive editor-level access.
- Viewers have read-only access to the presentation and all of its findings. Any authenticated user or user group can be added as a viewer. When a user group is added as a viewer, all members of that group receive read-only access.
Removing all editors and viewers (both individual and group) makes the presentation private again.
For presentations created by a scan, access is also granted through scan-level sharing. A user who is an editor or viewer of the scan gains the corresponding access to the scan's presentations, in addition to any sharing configured directly on the presentation. See Opportunity Scan — Sharing for details.
Slide Layouts
Slides come in two varieties: finding slides and text slides.
Finding Slides
Finding slides display one or more findings using a layout that controls how the findings are rendered. All layouts render each finding's description as annotation text below its content.
The supported layouts are:
chart— A single result displayed as a chart. Best used with one result per slide.chart-plus-single-values— The slide is split into two areas. Non-single-value results (bar, column, timeseries, or scatter charts) appear in the main chart area. Results withsingleValuechart type appear in a side column as key metrics.data-table— Each result on the slide is rendered as a data table.chart-with-associated-table— A single result is displayed as both a chart and a data table below it. Best used with one result per slide.side-by-side-charts— Two charts rendered side by side. Expects exactly 2 results, both chart types (bar, column, timeseries, or scatter). By default each chart gets 50% of the width, but per-resultwidthin chart instructions controls proportional sizing (e.g., one chart at 70% and another at 30%).single-value-grid— A grid of single-value KPIs. Expects 2–6 results, all withsingleValuechart type. Automatically arranges into 2 or 3 columns based on the number of results.stacked-charts— Two charts stacked vertically (top/bottom). Expects exactly 2 results, both chart types (bar, column, timeseries, or scatter).
If a finding slide has no layout set, it falls back to displaying result titles only.
Text Slides
Text slides contain markdown-formatted text with no findings. They are useful for title slides, executive summaries, section dividers, recommendations, or any narrative content that accompanies the data.
In the browser, text slides support the full range of markdown: headings, bold, italic, bullet and numbered lists, code blocks, tables, blockquotes, and horizontal rules. The PowerPoint export supports a subset — headings sized by level, bold, italic, and both list types.
Text slides and finding slides are mutually exclusive — a slide is either one or the other. To convert a finding slide into a text slide (or vice versa), delete the slide first and insert a new one at the same position.
Presentation Builder
Presentation owners and editors can use the Build button on the presentation detail page to open an AI-powered workspace for organizing slides. The builder helps arrange findings into slides with appropriate layouts.
How It Works
The builder interface has two panels:
- Left panel — A conversation with the AI, focused on presentation layout rather than database queries.
- Right panel — A live preview of the current slide layout, updated in real time as the AI makes changes. Each slide is shown with a skeleton rendering of its layout so you can see the structure before viewing the full presentation.
Available Tools
The builder's tools let the AI list and inspect the presentation's findings, assign them to slides, insert and delete slides, choose each slide's layout, write text slides, and restyle a finding's title, description, or chart. See the Tools Reference for the full list and the chart types each finding can be rendered as.
Two behaviors are worth knowing because they change slides you did not name. Moving a finding to a different slide removes its previous slide if that leaves it empty, and remaining slides are renumbered. Deleting a finding slide returns its findings to the unused pool rather than discarding them.
Interrupting the Builder
While the AI is working, the owner or an editor can click the Interrupt button to stop it mid-build. The AI stops at its next tool call boundary, acknowledges the interruption, and asks how you would like to proceed. Any layout changes already made before the interruption are preserved.
Persistence
Conversation history is persisted automatically. If you leave the builder and return later, the conversation resumes where you left off. Layout changes made by the AI are saved to the presentation immediately.
Snapshots
A snapshot is an immutable, point-in-time copy of a presentation. Snapshots capture the presentation's current layout and finding data so that the presentation itself can be refreshed with updated query results without losing the historical view.
Creating a Snapshot
Owners and editors can use the Create Snapshot button on the presentation detail page, naming the snapshot or accepting a default of the presentation name and the current date. The presentation's layout and all finding data (SQL, results, titles, descriptions, chart configuration) are copied into the snapshot, and changes to the presentation after this point do not affect it.
Viewing Snapshots
A presentation with snapshots shows a collapsible list of them near its vintage, each linking to its own detail page. The snapshot detail page displays slides and findings identically to the presentation, with two differences: there is no Build button, since snapshots are read-only and cannot be modified through the builder, and the back link returns to the parent presentation rather than the presentations list.
Owners and editors can rename or delete a snapshot through Edit Snapshot Metadata. Deleting is a soft delete — the snapshot is hidden but the underlying data is retained.
Snapshots support the same PowerPoint and Excel exports as presentations, subject to the same permission rules.
Refreshing Data
Over time the data behind a presentation changes, so the query results captured when each finding was added grow stale. Owners and editors can use the Refresh Data in Presentation button to bring the presentation up to date. Because a refresh overwrites the stored results, consider creating a snapshot first if you want to preserve the pre-refresh view. Only findings placed on a slide are updated; anything on the Unused Findings list is ignored.
Refreshing opens the builder and proceeds in two phases:
- Re-running queries. Each finding placed on a slide has its query re-run against the current data, replacing its cached results. If a query can no longer run — for example, because the underlying data model changed — that finding's results are cleared and its caption is set to "Data is no longer available for this finding." Every refreshed caption and text slide is flagged with a
[Pending Update]prefix, marking the wording as needing review. - Reconciling the narrative. Once the data is refreshed, the AI takes over automatically. It reviews each
[Pending Update]item, rewrites captions and text slides where the refreshed data changed the story, and simply removes the[Pending Update]prefix where the existing wording is still accurate. This kickoff instruction appears in the conversation attributed to Ursa Compass rather than to you.
The AI is not allowed to hand control back to you while any [Pending Update] item remains, so the reconciliation runs to completion before the builder becomes interactive again. Because the whole refresh runs through the builder, the slides and captions update live on screen as the AI works.
Exporting Presentations
Presentations can be exported as PowerPoint (.pptx) or Excel (.xlsx) files. The export buttons are available on the presentation detail page.
There are two export modes, depending on the user's permissions:
- Download — Users with Allow PHI Download rights see Download as PowerPoint and Download as Excel buttons that download the file directly to their browser.
- Export to bucket — Users without local download rights but with Allow PHI Export rights (and a configured export container) see Export to PowerPoint and Export to Excel buttons that write the file to the configured cloud storage bucket instead of downloading it.
Users who have neither permission see disabled buttons with a tooltip indicating they need to contact their administrator.
PowerPoint Export
The exported PPTX contains native PowerPoint charts and tables built from the underlying data — not screenshots or images. This means charts can be further customized directly in PowerPoint: changing colors, adjusting axis ranges, editing labels, or modifying the data series.
The PPTX is intentionally stripped of styling so that a PowerPoint theme can be easily applied. Charts reference the theme's accent colors rather than hardcoded ones, so applying a corporate theme automatically recolors every chart to match. Single-value KPIs are the exception — they use semantically meaningful colors (red, green, gold, blue, grey) that are preserved regardless of theme.
All finding slide layouts, chart types, and styling options (field formatting, height, width, annotations) are faithfully reproduced in the export.
Excel Export
The exported XLSX contains one worksheet per finding in the presentation layout, labeled alphabetically (A, B, C, ...). Each worksheet contains the raw query result data from that finding's data table. An Index worksheet is included with a row per finding listing its sheet label, title, annotation, and SQL query. Text slides are not included in the Excel export.
Editing and Deleting
Owners and editors can edit the name, description, tags, editors, and viewers from the presentation detail page via the Edit Presentation button. The edit form also provides a Delete Presentation option, which soft-deletes the presentation. Deleting a presentation does not affect the source findings.
Access Control
- Viewers can view presentation lists and details for presentations shared with them. Access to a presentation grants visibility to all findings it contains, regardless of whether the viewer has access to the source diagnostics.
- Editors and owners can create, edit, and delete presentations, add or remove findings, use the builder, and manage sharing.