How Appath Handles Your Data, Your Buyers’ Data and Its Own Security

• Every seller’s data is held under row-level security in Postgres: each table a seller owns carries the seller’s identifier, and the database itself refuses a read or write outside it.
• Appath never combines one seller’s data with another’s, gives no seller competitive intelligence derived from another seller, and publishes no insights about a sales channel’s business.
• A supplier sees its own products and the purchase orders it fulfils for each seller, and no statistic across the sellers it works with.
• Isolation is tested at the application’s HTTP boundary as well as in the database.

• A buyer’s and recipient’s details are used only to fulfil the order they came with: to route it, to show it to the seller’s own users, and to prepare it for the supplier it is routed to.
• A supplier receives those details, and the lines it fulfils, only after accepting Appath’s supplier terms, which limit their use to that order, and after Appath approves it. It never receives the seller’s cost or margin.
• Buyer data never reaches an AI model, and it is kept out of ordinary application logs and error reports.
• Amazon customer data is deleted on the schedule published in the Privacy Policy and on the Data Deletion page.
• Appath does not contact a seller’s buyers, uses no buyer-messaging service, and does not solicit or influence reviews, ratings or feedback.

The deletion schedule: Privacy Policy §09 and Data Deletion.

• Appath AI reads a seller’s data through fixed, deterministic tools inside that seller’s account. No tool can reach another seller’s data, and none calls a sales channel.
• Its figures come from those reads, each with its as-of time; a number in an answer that no read returned is marked before anyone sees it.
• It proposes and never acts: every suggestion opens a step a person confirms.
• Orders reach it without buyer details, on every channel, and the Assistant’s model calls are processed in the United States.
• As the Privacy Policy states, Appath does not use Amazon Information to train generalized or shared AI models, and does not use other customer data to train generalized or shared AI models without authorization.

More on how Appath AI works: Appath AI. The commitment in full: Privacy Policy §06.

• Encryption in transit and at rest.
• Vulnerability scanning on every change and every week: dependency, static-analysis, container-image and infrastructure scans, each a required check before code can merge.
• Application-level security monitoring, with alarms that notify the team, including an alarm for the monitoring itself going silent.
• An incident response plan, exercised in a tabletop.
• Account lockout after repeated failed sign-ins.
• An annual penetration test, most recently in August 2026, run internally and re-executed with industry-standard tools; its findings were fixed and retested, or declined with a recorded decision.

Appath is an approved Amazon Selling Partner API developer and holds the restricted Direct-to-Customer Shipping role, which Amazon grants only after its data-protection assessment, to applications that handle buyer data. Holding the role is an access approval, not an endorsement: Amazon does not endorse Appath.

Appath acts on a seller’s behalf, under that seller’s own authorization, and only within the permissions the seller grants through each sales channel. The seller remains the seller of record for every order.