Model and agent providers
Current listing path
Listing v0.1 uses deterministic parsing, mapping, and validation rules. Submitted product rows are not sent to GPT, an external model, or a third-party agent provider on this path. Owning model-token capacity does not change this boundary: content is sent only when an explicitly configured product version and provider contract allow it.
Future model-backed versions
All listing checks, model calls, and agent skills are currently free to users ($0.00) and subsidized by the project. Before any model-backed skill is enabled, we will publish the provider name, model or route, processing region, data-retention and training terms, sub-processors, and whether the provider supports deletion. The product will send only the fields allowed by that skill's data contract, record an opaque request reference and project cost, and avoid storing full prompts or responses in ordinary logs.
Where a provider retains data or transfers it across regions, the privacy notice and any required consent or data-processing terms will be updated before the change takes effect. Users may continue using the deterministic path where it is offered.
Human review remains required
Model or agent output is a draft. You must check factual support, prohibited claims, platform rules, prices, and local regulatory requirements before publishing. We do not treat a model confidence score as evidence that a claim is true.
Questions
Free-beta participants can ask about the active processing path through their invitation channel. Account owners can export or delete workspace data in the app. See the privacy notice for the current region and retention status. Any future provider or paid mode will require an updated notice before activation.