Trust Notes

Statements about a generated backend are bound to a specific artifact version. Validation evidence is produced by running that version in a sandbox, and a later edit produces a new version with its own evidence — an old passing run is not recycled as proof for changed source.

Deployment preflight reports missing prerequisites as structured errors. The platform does not return synthetic success URLs, and a generated artifact is never presented as a live public endpoint.

Staging and production are separate environments with separate evidence. A healthy staging deployment is not treated as proof that production is healthy, and neither is treated as proof about your own deployed runtime, which runs outside our control plane.

You retain the rights to your specs and generated code, and your content is not used to train shared models. Generated artifacts are downloadable as source so claims about the code can be verified by reading it. The full data-handling picture, including the current sub-processor list, lives in the Privacy Policy.

Model choices shown in the product use Axtra’s own model labels, resolved server-side. We do not expose raw provider identifiers or provider credentials at the customer boundary, which lets us route, upgrade, and retire underlying models without breaking your configuration.

We do not currently publish compliance certifications, customer outcome statistics, uptime SLAs, or numerical performance guarantees. If a page appears to make such a claim, treat it as an error and tell us at hello@axtl.dev.

The one time-bound commitment we do publish is a support response target: on Studio and Enterprise plans we respond to support requests within 24 hours on business days, as set out in the Terms of Use. It is a support commitment, not an uptime or performance guarantee, and Free and Builder support remains best effort.