Reports
Reports turn console analyses into documents that survive contact with a reviewer. Instead of screenshotting chat answers, the console captures analyses as structured reports: every number traceable to the tool run that produced it, every chart regenerable, every assumption written down. The workspace exists because an operational conclusion is only worth what its evidence trail is worth.
The reports workspace in fullscreen mode: section list on the left, an open report with a chart block and a calculation block, and the chart library drawer showing regen metadata.
What a report is
A report is a titled document of sections, each holding typed blocks rather than freeform text:
| Block | Contents |
|---|---|
| Markdown | Narrative text, editable in place |
| Chart | A reference into the chart library |
| Table | Titled columns and rows |
| Metric | A headline figure with its label and detail |
| Calculation | A titled calculation with its method, the worked math, and its assumption list |
| Asset reference | A link back to a scene entity |
Captured analyses arrive already organized: a summary of the conclusion, the findings, the supporting charts, the method (which tools ran, with what windows, steps, and masks), and provenance (what data, at what sim time, under which filters). The calculation block is the load-bearing one for review: it separates the method from the arithmetic from the assumptions, so a reviewer can attack each independently.
Capturing analyses
Any analytical answer in chat can be captured into a report, either into a new document or appended to an open one. Because insight cards already carry scope, window, and assumptions, capture is lossless: the report holds the same evidence the transcript showed, not a prose paraphrase. Reports also record which workflow and parameters generated them, so a weekly report is diffable against last week's run of the same template.
Editing
Reports are editable documents, not frozen exports:
- Markdown blocks edit in place; add narrative and operator notes between the evidence blocks.
- Blocks reorder within a section (move up and down), and sections can be retitled.
- Blocks and sections can be deleted; the underlying chart stays in the library.
- A fullscreen mode expands the workspace over the globe for sustained writing and review sessions.
Reports persist in the browser alongside your sessions and are listed newest first with their section and block counts.
The chart library
Every chart the assistant builds lands in a shared library, and each entry keeps its receipts:
- Origin: the tool and prompt that produced it, and which reports use it.
- Regen metadata: the exact tool call and arguments behind the chart, so it can be re-run as-is or with a different window, metric, or filter set.
A chart in a report is a reference into the library, not a pasted image. Regenerating an entry updates it from live tool output; a chart is never a picture whose provenance has been lost.
Workflow templates
Ten operator-grade templates turn recurring reporting into a parameterized run. Each template is a precise instruction naming the deterministic tools to execute, so two people running the same workflow get structurally identical reports.
| Workflow | Purpose | Key parameters |
|---|---|---|
| Daily fleet health | Posture, degraded assets, utilization and latency trends for shift handover | date, fleet |
| Weekly capacity and utilization | Where capacity is consumed, saturated, or stranded across the week | week start, fleet |
| Coverage comparison | Two constellation designs scored on identical windows, side by side | design A, design B, window, region |
| Gateway proposal | Customer-facing candidate site comparison with rain sensitivity | customer, candidate sites, objective |
| Link budget validation | One link budget walked end to end and reconciled with stored telemetry | link, required margin |
| Launch readiness review | Go/no-go checklist: injection orbit coverage, first contacts, TT&C geometry | mission, launch date |
| Anomaly investigation | Evidence-driven timeline and root-cause analysis for one asset | asset, window |
| Executive dashboard | Leadership summary: headline metrics, trends, incidents, decisions | period |
| Regulatory coverage | Reproducible service-availability evidence for a country filing | country, window |
| Orbital trade study | N constellation configurations ranked on one metric with identical scoring | configurations, metric |
Required parameters are prompted before the run; optional ones default sensibly and the report states the default that was chosen. The templates encode the console's evidence discipline: state every window and mask, mark synthetic runs as synthetic, report truncated coverage plainly, and never infer a cause the numbers do not support.
Exports
| Format | Behavior |
|---|---|
| Markdown | Portable text with tables; charts export as their data |
| HTML | Standalone styled document |
| Print-based: a print-optimized rendering opens in the browser's native print-to-PDF path, with charts rendered as data tables for archival fidelity |
The PDF path is deliberately dependency-free: nothing is rasterized server-side, and the printed document carries the same numbers as the live report.
Common workflows
| Task | How |
|---|---|
| Start a shift handover | Run Daily fleet health with today's date |
| Preserve an incident analysis | Capture the chat investigation, then annotate with markdown notes |
| Refresh a stale chart | Regenerate it from the library with a wider window |
| Prepare a customer artifact | Gateway proposal workflow, export to PDF |
| Compare designs for a review board | Orbital trade study, one metric, identical scoring stated once |
Example scenarios
A filing that survives scrutiny. A regulatory affairs engineer runs the Regulatory coverage workflow for Norway over 24 hours. The report states the achieved availability with its exact evaluation window, step, and elevation mask, the station inventory used, and each gap's duration, and notes the evaluation is deterministic. Six weeks later, the regulator's technical reviewer asks how a number was computed; the calculation block already answers.
The weekly report that diffs. An ops lead runs Weekly capacity and utilization every Monday with the same parameters. Because the workflow and parameters are recorded on each report and every figure is tool-traceable, week-over-week deltas are real comparisons of identical computations, not two analysts' differing methods.