Where your data flows for inference and 3D generation (OpenRouter, Meshy, Railway)

When you build with HIIE, some requests are sent to third-party and self-hosted services to do the actual work. Here's exactly where your data goes, documented on our AI Transparency page.

AI inference: OpenRouter

By default, prompts for managed inference are routed to large language models through OpenRouter, which in turn routes to the underlying model providers. Managed inference is metered with service credits. If you bring your own provider key (BYO), the model runs on your own account and is unmetered; your BYO key is encrypted at rest and never returned to the browser.

3D generation: Meshy and Railway

3D generation requests go to one of two places depending on the path:

• The premium path sends your text prompts (text-to-3D) and concept images (image-to-3D) to Meshy (api.meshy.ai), which generates the textured mesh and returns the model files.
• The default engine runs on a self-hosted build123d CAD service hosted on Railway, which generates parametric B-rep geometry.

Both produce real model outputs that are archived back into your project.

Data at rest stays encrypted

Regardless of where inference or 3D work happens, your per-user data is encrypted at rest with server-only keys. Data in transit uses TLS.

Policy on third-party use

HIIE states as a policy commitment that your data is not sold and not used to train third-party models. Note this is a policy commitment about how data is handled by HIIE and its vendors, not a cryptographic guarantee — your prompts, concept images, and CAD inputs are processed by these providers to deliver the result you asked for.

You can read the full account, including a markdown mirror at /ai-transparency.md, on the AI Transparency page.