AMALYZE

Transparency

AI & Amazon Data Use

AMALYZE helps Amazon Sellers and Vendors work faster on their own catalog, ads, and listings. This page explains how we use AI models and how we handle data retrieved from Amazon's SP-API, Advertising API, and public product pages, in line with the Amazon Services API Acceptable Use Policy (AUP, section 2.4 on AI transparency and section 4 on data usage).

How we use AI

  • Listing audits and drafts. We use large language models (currently frontier models from OpenAI, Anthropic, and Google, accessed through their commercial APIs) to score listing quality, suggest improvements, and draft titles, bullets, A+ copy, and translations.
  • Keyword and search-term analysis. We use embeddings and classical NLP to cluster keywords, deduplicate variants, and detect intent — all scoped to the seller's own catalog and ad accounts.
  • Ads automation. Bid, budget, and negation rules run on deterministic logic against the seller's own campaign data. Where AI assists (e.g. anomaly explanations), the model output is labelled as a suggestion, not an automated write-back, unless the seller explicitly enables it.

Accuracy and data freshness

AI-generated outputs are drafts and suggestions, not guarantees. Language models can be wrong, especially on numeric reasoning and long-tail claims. Every AI-drafted listing change stays a draft until the seller reviews and publishes it. Underlying data is refreshed on the cadence Amazon provides — typically hourly for reports we poll, near-real-time for Advertising API metrics, and on-demand for public product pages. Freshness of each dataset is shown next to the metric in AMALYZE Shield.

What we do not do with your data

  • No training on your Amazon data. We do not train, fine-tune, or otherwise improve any AI model on data retrieved from Amazon on your behalf — not our own models, and not the third-party model providers we call. API calls to those providers are made with the “no training” / zero-retention settings their enterprise agreements offer.
  • No cross-seller aggregation. We do not aggregate SP-API data across Authorized Users to provide, sell, or benchmark against other sellers. Each seller's data is used only to support that seller (AUP 4.4, 4.6).
  • No resale of Amazon data. We do not offer or promote external data services that vend information retrieved from Amazon (AUP 4.3).
  • No PII targeting. Customer PII from the SP-API is used only for the merchant-fulfilled and tax/regulatory purposes Amazon permits (AUP 4.1, 4.2).

Public vs. seller-authorized data

AMALYZE distinguishes two data sources. Publicly visible product-page fields (title, images, price, review count, BSR) are used to support the seller's decisions on their own listings and categories. Seller-authorized SP-API and Advertising API reports are pulled only after the seller grants access, are stored per-seller, and are never combined with another seller's authorized data.

Contact

Questions about this disclosure? Reach out via the details on the homepage. Related reading: BSR.

Last updated: July 2026