Inventory Management
NetSuite Unknown
2026-07-24

Item Creation Assistant: AI-Driven Bulk Inventory Item Creation from PDFs

NetSuite adds an Item Creation Assistant that uses generative AI to extract product data from uploaded PDFs and bulk-create Inventory Item records, limited to 100 items per upload with PDF-only input.

Affects:Inventory Item recordAdvanced Tax EngineMultiple CurrencySubsidiariesCSV/Import framework

What changed

A new Item Creation Assistant page is available in the UI. It accepts a PDF upload (product catalog, price list, etc.), runs it through a generative-AI extraction pipeline, and presents parsed item data for review before creating records in bulk.

Constraints

  • Supported item type: Inventory Item only — no Assembly, Non-Inventory, Service, Kit, or other item types.
  • File format: PDF only, max 50 MB.
  • Extraction cap: 100 items per upload. Larger catalogs must be split manually.
  • Files must contain structured product data (names, descriptions, prices). Large or unstructured PDFs may fail processing.

Extracted fields

  • Item name/number
  • Item type
  • Tax schedule (visible only when Advanced Tax Engine feature is enabled)
  • Subsidiary (visible only when Subsidiary Column feature is enabled)
  • Display name/code
  • Description (truncated to 200 characters)
  • UPC code
  • Purchase rate and purchase price currency (visible when Multiple Currency feature is enabled)
  • Sales rate and sales price currency (visible when Multiple Currency feature is enabled)

If the selected currency does not match the account's base currency, the purchase price is automatically converted to the base currency at item-creation time.

Navigation paths (role-dependent)

  • Administrator: Lists > Accounting > Item creation assistant
  • Controller: Financial > Lists > Item creation assistant
  • Inventory & Purchasing: Inventory > Lists > Item creation assistant
  • Sales: Customers > Transactions > Item creation assistant

Post-creation workflow

After clicking Create Items, results appear on a Processed Records page. Each row shows a Line Status and Result column. Rows marked Success link directly to the created item record. Rows marked Failed include an error message — the source PDF must be corrected and re-uploaded to retry those items. There is no partial-retry mechanism described; the entire file must be re-uploaded.

Technical considerations

  • No API or SuiteScript surface documented. Oracle's help page describes a UI-only workflow. There is no mention of a REST endpoint, SuiteScript module, or SuiteTalk operation for programmatic access to this assistant. If your workflows depend on automated item creation, this feature does not replace existing approaches via CSV Import, N/record, or the SOAP/REST record APIs.
  • AI accuracy caveat. Oracle explicitly warns that extracted data may be inaccurate or incomplete. Every field should be reviewed before committing. This is not a hands-off bulk import tool.
  • The 200-character description limit is tighter than the standard Item record description field. Descriptions longer than 200 characters in the source PDF will be silently truncated.
  • The feature appears to use the existing CSV/Import infrastructure under the hood (the View Import Process link and Processed Records page are part of that framework), but Oracle does not confirm this explicitly.

What to do

  1. Evaluate fit. This assistant is useful for one-off or periodic catalog ingestion where the source is a vendor PDF. It is not a replacement for scripted or integration-driven item creation.
  2. Enable prerequisites. If you need tax schedule, subsidiary, or multi-currency fields during extraction, confirm that Advanced Tax Engine, Subsidiary Column, and Multiple Currency features are enabled in your account.
  3. Prep your PDFs. Ensure files are under 50 MB, contain structured tabular product data, and are split to stay under the 100-item extraction limit.
  4. Assign roles. Verify that users who need access have one of the listed roles (Administrator, Controller, Inventory & Purchasing, or Sales) or a custom role with the equivalent menu path.
  5. Review every record. Treat AI-extracted data as a draft. Validate item names, UPC codes, and pricing before clicking Create Items — corrections after creation require manual edits to each item record.
  6. Monitor the Processed Records page. Check for failed rows and address source-file issues before re-uploading.