HardwareHub Workbench
Data + IP
Before using Workbench, review how project information is handled. This is a practical product policy summary, not legal advice. External beta users should also review the beta terms and privacy notice below.
External beta terms
Workbench external beta is an invite-only, experimental product. AI output can be incomplete or wrong, and HardwareHub work does not start until a user explicitly approves a Work Request or other reviewed handoff.
- Users must keep other customers' information, beta details, and HardwareHub non-public workflow details confidential.
- Users are responsible for reviewing AI suggestions, supplier notes, dimensions, materials, timelines, and work requests before relying on them.
- Feedback, issue reports, and usage notes can be used by HardwareHub to improve Workbench and HardwareHub services.
- The beta is provided without a production service-level commitment, warranty, or guarantee that any product will be manufacturable, safe, compliant, or ready for sale.
- Support is limited to the beta support path HardwareHub provides to the invited user or company.
External beta privacy notice
HardwareHub uses Workbench project data to operate the beta, answer project questions, build project records, improve the product, and support invited users.
- Project data includes project briefs, questions, answers, timelines, assemblies, parts, vendors, file metadata, uploaded files if uploads are enabled, and support messages.
- When live AI is enabled, Workbench sends bounded project context to the selected AI provider only after access, usage-meter, and cost gates pass.
- Provider memory is not the source of truth. Workbench project records, approved memory, audit records, and file versions are the source of truth.
- Archive, export, restore, and secure delete are separate lifecycle actions. Delete is not enabled until the secure deletion path is reviewed and can be audited.
- HardwareHub does not sell customer project data. Local or private deployment remains an inquiry path for teams that cannot use cloud AI.
Upload content handling
Workbench file uploads are enabled for approved Workbench projects after project access checks. Files are stored as private, project-scoped Workbench file versions. File access uses short-lived server-issued links after authorization, and uploaded content is not sent to an AI provider unless HardwareHub separately enables and verifies a file-content AI path.
- Accepted files are stored in HardwareHub's private Workbench storage bucket and are scoped to the project where they were uploaded.
- Raw storage paths are not returned to the browser; file access uses server-issued URLs only after authorization checks pass.
- Upload finalization must verify storage metadata before a file version is recorded.
- Archive and restore do not delete uploaded files. Secure delete remains disabled until the deletion path is separately reviewed and audited.
Delete content handling
Workbench secure delete remains disabled until HardwareHub separately verifies the checked deletion request path, storage cleanup, authorization, and audit evidence. Archive, restore, export, and delete are separate lifecycle actions.
- Archive hides a project from active views but does not delete project data.
- A delete request is not a hard delete unless the secure deletion path is enabled and audited.
- Storage-object deletion stays service-controlled and is not triggered by the external beta UI while deletion is disabled.
You own your project information.
HardwareHub does not claim ownership of your ideas, product plans, files, designs, requirements, supplier notes, or project data. Your product is yours.
AI engine provider policy
Current provider: OpenAI API or Anthropic API.
Workbench may use the selected external AI provider to generate responses. HardwareHub keeps project memory, files, account details, and pricing rules in Workbench, then sends the needed context to the selected provider at runtime.
Provider policies can change. HardwareHub should keep this page updated before launch, whenever provider settings change, and before any live AI provider is enabled.
Need a local or private deployment?
If your company cannot use cloud AI for sensitive product data, contact HardwareHub about a local or private-network version of the AI engineer workflow for your team, machines, and internal project data.
Ask about local deploymentPersisted beta acceptance path
When WORKBENCH_BETA_ACCEPTANCE_REQUIRED is enabled, Workbench records in-app acceptance for the active terms, privacy notice, Data + IP summary, upload content handling, and deletion content handling versions before a beta user can proceed into Workbench.