The Insights Sandbox

2 min read

The Insights Sandbox lets you build an insight and try it on the layer you’re looking at right away — before committing it to your catalog. It’s the fast lane for experimentation: tweak an expression, apply it, see the result on the Globe, adjust, and only save the ones that earn their place.

Where it is #

The Sandbox is a collapsible Insights Sandbox section inside the workflow result panel — expand it once you have a Visual on the Globe. (It’s a Remote Miner feature.) A banner reminds you to have an active layer selected so the sandbox insight has something to apply to.

📷 Screenshot pending — the Insights Sandbox section expanded under a run card.

Building and testing #

  1. Pick an Insight type — the same types as the full builder (Band Ratios, Wavelength Analysis, PCA, Distance, Mixture, Matched Filtering, Scripted), filtered to what your instrument supports. On a thematic instrument the sandbox tells you to switch to a spectral instrument instead.
  2. A modal opens with that type’s editor; configure it and hit Create (or Update when editing one you already made).
  3. Your creations collect in the Sandbox insights list. Each shows a compatibility indicator — it flags (and disables) any insight that doesn’t fit the current instrument.
  4. Select one and hit Apply to render it over the active layer. Iterate as much as you like.

Saving the keepers #

A sandbox insight is temporary until you save it. On any item:

  • Save Insight promotes it into your real catalog — a dialog asks for a Name (at least 3 characters), a Category, and an optional Description. Once saved, it shows a green “saved” check.
  • Edit Insight reopens its editor; Delete Insight removes it. Deleting an unsaved sandbox insight just drops it from the list; deleting a saved one asks you to confirm, because it permanently removes the catalog insight.

When to use it #

  • Dialing in a threshold or expression — apply, eyeball the map, adjust, repeat, without cluttering your catalog with half-baked attempts.
  • Comparing a few candidate indices on the same scene before deciding which to keep.
  • Teaching/testing — try an idea end-to-end in seconds, then save it as a proper insight (or a Stack) when it works.
Updated on July 27, 2026