Hosted data handling

Staged approach

1. A local demonstration with synthetic files (no upload).
2. A private, authenticated trial with synthetic data only.
3. An authorized customer pilot with written authorization and controlled retention.
4. A public trial only after an external security review and counsel-approved privacy and data-handling pages.

What would be uploaded

Only what you choose to upload, after a screen that names the file, its data classes (addresses, device inventories, control configuration, alarm configuration, redacted secrets), the storage region, and the deletion date. The trial would be read-only and would never require control permissions on a building.

How it would be handled

• Size caps and content checks on intake; the same importer caps as the edge.
• Malware scanning of the archive before parsing.
• Parsing in an isolated job: separate container, no network, read-only root, memory and CPU limits, timeout.
• Storage encrypted with a per-tenant key; tenant isolation enforced server-side on every query.
• The raw upload deleted after the job unless you opt to retain it.
• An access log of every read of your data by project staff, available to you.

Retention and deletion (proposed)

• Trial data deleted 30 days after last activity.
• Pilot data per the pilot agreement.
• Deletion on request within 7 days, including backup rotation and logs that could identify the site.

Subprocessors (proposed)

The hosting provider, object storage, and transactional email. Named on this page before launch. No AI providers. No analytics providers.

Never

Customer engineering data is never used to train models. No dark patterns, no surprise billing, no silent transfer to anyone.