August 2026 release notes for Inventory Management in Automation Cloud.

The endpoint for adding a custom object to a solution is now POST /api/v2/schema/add-object, replacing POST /api/v2/schema/custom-table. The new name pairs with POST /api/v2/schema/{objectName}/add-attribute — one adds a table, the other adds a column. Request and response bodies are unchanged, and the previous custom-table path continues to work, though it is deprecated and may be removed in a future release; add-object is the recommended alternative going forward. See Add a custom object.

The Add a custom object payload reference previously listed primaryKeys as optional. It has always been required — a request without a non-empty primaryKeys array is rejected with 400 Bad Request, and every name in it must match a column declared in columns.

You can now ingest a large dataset by uploading a CSV (or .csv.gz) file directly to an existing table, instead of sending rows as JSON. Uploads are scanned for malware, validated against the schema per row, and processed asynchronously — poll the returned requestId for status and success/failure counts. See Bulk CSV upload.

The Data Quality Dashboard now surfaces object-level validation results alongside the existing row-level metrics — freshness age vs. threshold, missing-data gaps, and per-period shortfall, with per-table drill-down. See Object-level validations.