Skip to content

Data handling and security

Your PDF is not passed to OpenAI or Anthropic. The models LevelOps uses run inside its own Azure environment, so your documents are never sent to OpenAI the company.

  1. Your PDF arrives in a dedicated order inbox and is processed automatically. All data is encrypted in transit and at rest. The pipeline runs in LevelOps’ own Azure environment.

  2. The document is first converted into structured text by a specialized document-parsing service that is SOC 2 Type II certified. Under its terms, your document is never used for model training, is encrypted in transit and at rest, and is automatically and permanently deleted within 48 hours of processing. LevelOps shares its full sub-processor list on request.

  3. The extracted content is then interpreted by OpenAI models running inside LevelOps’ own Azure environment (Azure OpenAI). Nothing is sent to OpenAI the company. Microsoft does not use the data to train models. It is not visible to any other customer. Microsoft retains processing data only for abuse monitoring, for a maximum of 30 days, after which it is deleted.

  4. On LevelOps servers, storage is transient and limited to processing. The source document and its raw extracted content are automatically deleted within 48 hours. Short-lived processing logs are purged after 2 to 3 days. The output, the draft order, is created directly in your own Shopify store.

  5. LevelOps does not sell your data, and does not use your documents to train AI models, its own or anyone else’s. The only learning in the system is scoped to your account: your corrections and product mappings improve extraction accuracy for you. No customer’s data is visible to another customer.

AI extracts order data from the document, then matches products to your Shopify catalog. SKU lookup runs first. If no SKU is present, semantic matching compares the product description to your catalog. See How product matching works.

The draft order is created in your Shopify store. The original purchase order is saved to your Shopify Files and linked from the draft order. Copies used for processing on LevelOps servers are deleted within 48 hours.

No. LevelOps does not use your documents to train AI models, its own or anyone else’s. The document-parsing service does not use your document for model training. Microsoft does not use the data to train models. The only learning is scoped to your account: your corrections and product mappings improve extraction accuracy for you.

No customer’s data is visible to another customer. Processing data is not visible to other Microsoft customers. LevelOps shares its full sub-processor list on request.

The source document and raw extracted content are deleted from LevelOps servers within 48 hours. The parser deletes its copy within 48 hours. Processing logs are purged after 2 to 3 days. Microsoft retains processing data for abuse monitoring for a maximum of 30 days. The draft order stays in your Shopify store.

Open a support ticket or email support@levelops.co. Processing copies are also deleted automatically on the timelines above. Draft orders and files in your Shopify store stay under your control.