Dashboards & reports
Narrato has two artifact types the assistant can build for you, and the difference matters:
| Dashboard | Report | |
|---|---|---|
| What it is | A set of widgets, each backed by a query | A written document with numbers inline |
| Freshness | Live: every view re-runs the queries | Frozen: the numbers are from when it was written |
| Good for | Watching something over time | Answering “what happened?” once, shareably |
| Changes when your data changes? | Yes, automatically | Never (that’s the point) |
Dashboards: live lenses
Section titled “Dashboards: live lenses”Ask in chat (“build me a dashboard of activation funnel health”) or use the guided first-run setup. The assistant inspects your live event schema, writes each widget’s query, executes it against your real data before saving (a widget that errors or returns nothing gets fixed or dropped, not persisted), and picks sensible chart types.
Dashboards re-query on every view, with per-widget error isolation: one broken widget renders as an error card while the rest of the dashboard loads normally.
Reports: frozen snapshots
Section titled “Reports: frozen snapshots”Ask for a report when you want a document: “write up last week’s launch performance.” The assistant runs the analysis, writes markdown with the numbers inline, and saves it. Reports never re-run. Six months later they still say exactly what they said, which is what you want from a record.
Every report carries a grounding receipt: the actual queries executed during generation, attached to the report so you can audit where each number came from. The assistant is mechanically prevented from saving a report when it ran no queries. No data, no report.
Where they live
Section titled “Where they live”Dashboards and Reports pages in the console list everything in your current workspace. Both belong to the project they were built from, so team members see the team’s artifacts and personal ones stay personal.