A data-management heavy week: connections got smarter sync across the board, the dashboard now sorts by last-edited project, and dataset updates gained a row-deletion option. The field app also picked up social sign-in.
Features
Inward (negative) buffer for buildable-area calculations
Buffer tools now support an Inward direction — shrinking a polygon rather than expanding it. This is useful for calculating buildable areas that exclude a setback distance from the boundary. The Outward/Inward toggle is available in the Standard Buffer function, the Multiple Buffer tool, and the Workflow buffer node. The live preview updates to reflect the chosen direction before you run the operation.
Social sign-in in the field app
Field app users can now sign in with Google, Microsoft, or Apple — the same providers available on the web. The sign-in page shows icon buttons below the existing email/password form. Users who registered with a social provider on the web no longer need a separate password to access their projects on mobile.
Projects sorted by last edited
The dashboard now shows "Edited X ago" on project cards and sorts projects by when they were last changed — the same behaviour as Figma and Google Docs. Editing a layer, element, interface, form, or any other project resource bumps the timestamp. Brand-new projects show "Created X ago" until they receive their first edit. Comments and form responses do not count as edits.
History panel
The Activity panel has been renamed to History and moved into the right side panel (next to Comments). It can be opened from the layer ellipsis menu or the datatable header in editor mode. The panel follows the selected layer and closes when no dataset is selected.
Smart merge sync for database connections
Postgres, BigQuery, Snowflake, Databricks, MySQL, MSSQL, and Oracle connections can now sync by merging rows rather than replacing the entire dataset on every sync. A unique key column and duplicate strategy can be configured when importing the table. Rows deleted in the source are soft-deleted in Atlas; rows that existed before the sync retain any geometry or annotation added in Atlas.
Native record ID and editable columns for Airtable, ClickUp, and Smartsheet
Airtable, ClickUp, and Smartsheet connections now capture the native record ID from each provider and use it as the merge key by default. This means smart sync is on automatically for these connections — without needing to choose a key column manually. Because row identity is now preserved across syncs, you can add new columns to a synced dataset and have them survive the next sync. Google Sheets is excluded (it has no stable record ID).
Delete rows missing from a file update
The dataset update flow (uploading a new file to replace or merge into an existing dataset) now includes a "Delete rows missing from update" option. When enabled, rows in the dataset that have no match in the uploaded file are removed. A red segment in the preview ledger shows how many rows will be deleted before you confirm. The default remains to keep unmatched rows, so existing uploads are unaffected.
Fixes and Improvements
- Map export filters — exported map images now reflect the active filtered view instead of showing the full unfiltered dataset
- Raster export legend — raster layer legends now appear correctly in map exports
- Navi status line — the thinking indicator now shows what Navi is actually doing ("Fetching traffic data") instead of rotating placeholder text
- Navi chat speed — fixed significant keystroke lag in the Navi chat input, especially noticeable with long conversation histories
- History completeness — datasets created from connections, analysis operations, or copy actions now appear in the project History
- Interface layer reactivity — checking a folder in the interface layer manager now takes effect immediately without needing to re-enter the interface
- Timeline and clustered markers — marker and cluster layers no longer disappear during timeline playback
- Google Sheets tab selection — fixed an issue where sheets with tab names longer than 31 characters always imported the first tab instead of the selected one
- Column conflict on re-upload — fixed a naming collision with id/id_2 columns when re-uploading a dataset
- Header-only CSV upload — column structure is now preserved when uploading a file that contains headers but no data rows
