Privacy

On behalf of a customer, and only to operate the service for them: supplier manifests and stocklists; catalog and product records; inventory movements, grades and repair history; order and settlement records from marketplaces the customer connects; and the account details of the people they invite — an email address and a password they choose themselves.

We do not ask for, and have no use for, a customer's end-buyer personal information.

Where a customer connects a marketplace or logistics account, we act on their authorisation and under that marketplace's own terms. Access is read-oriented: we ingest orders, settlements and catalog data to price and reconcile.

We do not scrape. Data reaches us from licensed and official sources only — files a customer gives us, and APIs used under their published terms. That is a standing rule in our engineering decisions, not a preference.

We use language models for one job: identity. A supplier calls a thing one name and a catalog calls it another, and the model proposes which catalog entry a vendor's description refers to.

What is sent to a model provider is a product description and a list of candidate product names. Not buyer information, not orders, not settlements, not prices, not account details. The request is one question with a numbered list, and the answer is one choice from that list.

We do not train models on customer data, and our model providers do not train on data submitted through their APIs under their current commercial terms. A model never sets a price: every figure that costs money is computed deterministically by our own software and can be traced to the inputs it came from.

• DigitalOcean — hosting and the managed database. Receives everything, because it is where the service runs.
• OpenAI and Anthropic — model providers for identity matching. Receive only the product descriptions and candidate names described above.

That is the whole list. There is no analytics service, no advertising pixel and no third-party tracker anywhere in this product — the only outbound requests it makes are to the two model providers above.

We do not sell data, we do not share it for advertising, and we do not use one customer's data to price another customer's inventory.

Traffic is encrypted in transit. Credentials and connection strings live on the server and are never committed to source control. Sign-in passwords are stored as salted hashes, never in readable form, and are chosen by the person themselves rather than issued to them.

We would rather describe controls we have than imply ones we do not. If a specific assurance matters to you before you connect an account, ask us and we will answer precisely.

We keep customer data for as long as the customer has an account with us, and delete it on request. Some records are written to be permanent for the product to work at all — the history of a decision, and what a unit actually returned, cannot be reconstructed once discarded — so a deletion request removes the account and its data rather than editing history in place.

A customer may ask what we hold about them, ask for a copy, ask us to correct it, or ask us to delete it. Revoking a marketplace connection stops further ingestion immediately.

If this policy changes materially we will say so here and update the effective date above. Questions, requests, or anything you want answered before connecting an account: dev@renewedai.com.