Overview
An Insight is a factory-scoped dashboard — a named page that holds one or more chart panels. Insights are independent of any specific simulation run, experiment, or Monte Carlo request, so they’re useful both for ad-hoc exploration (e.g., profiling uploaded data) and for durable dashboards your team navigates back to over time. Each Insight has a name and a description, both shown at the top of the dashboard. Under the hood, an Insight is a two-level structure: the dashboard lives at the top level (with name and description), and each chart is its own nested file beneath it. You can have many charts in one Insight, and Insights live alongside each other in the factory.When to Use Insights
Different surfaces in ProDex serve different analysis needs:- Insights — durable, navigable dashboards that aren’t tied to a single run. Use for cross-cutting views, profiling raw uploaded data, executive dashboards, anything you’ll come back to repeatedly.
- Run Results — charts and KPIs scoped to one specific simulation run. Use for inspecting that run’s behavior in detail.
- Experiments — side-by-side comparison of multiple Snapshots. Use when the question is “what’s different between these scenarios?”
- Monte Carlo — distributions and confidence intervals over many replications of one Snapshot. Use for statistical reasoning about a single configuration.
Creating an Insight
Click + New insight in the Insights view. Give it a Name (required) and a Description (optional, but recommended — a substantive description helps your team understand what they’re looking at when they open it later). You can optionally pick a Pipeline from the dropdown to associate the Insight with one of your data pipelines, or choose No pipeline.The Pipeline dropdown is a UI-only convenience. It helps with filtering and gives Dexter context, but the association isn’t persisted in the underlying data and doesn’t scope Dexter’s data access — Dexter still pulls from the same factory-wide
uploads/ and pipeline outputs regardless of the choice.Adding Charts
There are two ways to add a chart to an Insight, both of which route through Dexter:- From the Insight page — click the + Add Chart card next to the existing charts. A Create chart dialog opens where you describe the chart you want (e.g., “throughput by entity type, last 30 days”) and optionally pick a chart type. Dexter computes the data and renders the panel.
- From chat — ask Dexter directly in the conversation panel (“Add a histogram of cycle times from the latest run to my Q3 dashboard”). Same result, no dialog.
Chart Types
The Create-chart dialog supports the following chart types — all first-class options:| Type | Use when… |
|---|---|
| line | Tracking a metric over a continuous axis (time, position) |
| area | Same as line but emphasizing magnitude under the curve |
| stacked_area | Stacked area showing component contributions over time |
| bar | Comparing values across discrete categories |
| stacked_bar | Bars decomposed into stacked sub-categories |
| pie | Part-to-whole breakdown |
| scatter | Showing the relationship between two variables |
| histogram | Distribution of a single variable across bins |
| table | Tabular layout for dense numeric data |
| heatmap | Two-dimensional intensity grid |
| gantt | Time-based intervals for jobs or events |
| box_plot | Spread, quartiles, and outliers across groups |
| mixed | Combination chart (e.g., bars + a trend line on the same axis) |
Editing and Reordering
Ask Dexter to update an existing chart (“change the date range on the throughput chart to last 90 days”) and it’ll modify the chart in place. To reorder charts, ask Dexter to move them — the page renders charts in their stored order in the dashboard file.Refreshing Data
Charts are static snapshots of the data at generation time. When the underlying data changes (a pipeline re-runs with fresh inputs, or you upload new files), existing charts don’t update automatically. Ask Dexter to regenerate the affected charts — “refresh the throughput chart with the latest pipeline run” — and Dexter recomputes the data and replaces the snapshot in place.Sharing and Reporting
Insights are factory-scoped — anyone with access to the factory sees the same Insights. There’s no per-Insight permission layer; visibility follows the factory. Because chart data is committed inline in each chart’s JSON, Dexter can pull it directly when building a report. Ask Dexter for a deliverable — “build me a PDF of the Q3 dashboard for the operations review” — and Dexter composes a report saved underreports/ (PDF rendered via Typst, or an Excel export with the underlying data tables).