Staging draft Work in progress · For review; coverage and acceptance incomplete

Assess Private Cloud readiness and obtain help

Private Cloud remains disabled in the inspected configuration. This guide helps members, administrators, support staff and decision-makers understand that status and prepare a future assessment. It does not instruct anyone to activate the capability.

The frontend is in progress. Read-only inspection of 06 Final Screens established commercial legacy-review screens and, in the earlier governance pass, the disabled HiveAI boundary. It did not establish a complete private AI activation journey. The steps below are review and support responsibilities, not invented application controls.

Before you begin

Identify the question you need to answer: an unavailable HiveAI capability, a commercial contract review, an information-handling requirement or a future deployment assessment. Record the affected workspace and product context.

Keep confidential content and credentials out of general requests. Use safe record references and a short description. A subscription label, signed commercial arrangement or administrator role must not be treated as permission to activate private AI processing.

If you see that Private Cloud is disabled

Read the availability statement as a current limitation. Refreshing the page, changing a person's role or buying another plan does not demonstrate that the private AI deployment has been approved and installed.

Ask your workspace administrator or support owner whether your task can use an already available, authorized capability. Do not move protected content to another AI service as an outage workaround. An alternative processing boundary needs its own approval.

Keep any original conversation or processing reference if work has already started elsewhere. The disabled status does not establish that previous work or charges have been cancelled.

Distinguish commercial review from AI readiness

Platform staff may encounter Private Cloud Review within the commercial quote and contract experience. The inspected populated design lists existing organizations and review reasons; the empty design indicates that no matching legacy commercial items need review.

Those screens support commercial follow-up. Opening a review item does not install a model, change a workspace's hosting or enable the HiveAI private capability. Follow the commercial guide for the intended task and its implementation limitations.

If a request mixes commercial terms and AI readiness, assign both the commercial owner and the AI deployment/governance owner. Keep each decision and its supporting evidence clear.

Describe the business need

Prepare a short statement of the intended work, intended users, information categories, required processing location and the reason the currently available arrangement does not meet the need.

Include the expected workload, critical periods, support expectations and consequences of interruption. Explain which sources must be searchable and whether particularly restricted information is involved. These inputs allow the responsible owners to assess the requirement; they do not establish technical feasibility or approval.

Managers and executives should ask for a plain-language account of what will be supported, what remains unavailable, who will operate it and how success will be demonstrated.

Assign accountable owners

Agree who owns each responsibility before evaluating readiness:

Responsibility Decision or evidence needed
Deployment and AI-system ownership Exact intended use, boundary, users and accountable organizations
Model and embedding suppliers Approved services, immutable versions, terms and current evidence
Inference and database operations Installation, capacity, identities, monitoring and maintenance
Security, privacy and legal review Information handling, applicable obligations and supplier-role decisions
Incident and corrective-action ownership Reporting, containment, suspension and reapproval
Backup, recovery and retention Recovery objectives, restoration checks, deletion and legal-hold handling
Evidence custody and monitoring Current records, expiry review, changes and ongoing oversight

One organization may hold several responsibilities, but each must have an identified owner. A role that is conditionally unnecessary needs a recorded reason. Do not substitute a general “IT approved” statement for the required decisions.

Review the proposed boundary and evidence

Ask the governance and technical owners to demonstrate that the proposed chat, embedding and search components belong to the same reviewed boundary. Their versions, information ceilings, region, evidence and operating terms must be compatible.

Review actual supplier and deployment evidence, including when it expires and who must renew it. A configured date or reference is not independent certification. Use the governance guide for policy and material-change responsibilities.

A current decision must cover the intended use. It must not be inferred from approval for a different service, purpose, region or information category.

Separate configuration from a successful rehearsal

Readiness evidence should answer several different questions:

  1. Does the current approved policy permit this deployment and capability?
  2. Does the installed configuration match the approved profile?
  3. Can the intended user complete the intended action under current permissions?
  4. Are the actual provider and database services working?
  5. Are search results authorized, current and supported by usable citations?
  6. Can failures, outstanding charges, deletion and restoration be recovered correctly?

A process-health indicator answers only the checks it actually performs. The current implementation does not establish all of these outcomes. In particular, a disabled-component result is not an availability test, and a support-search check is not proof that every workspace source has been processed.

Plan source preparation and controlled availability

Agree which authoritative sources may be used and who approves their inclusion. Preserve their permissions, classifications, versions and removal obligations. Preparing search data must not make unpublished or unauthorized information available.

The implementation separates preparation, verification and activation of a search generation. The future qualification must prove the complete workflow, including failed preparation and withdrawal. Current private support-publication and workspace-manual publication gaps remain unresolved.

Do not ask members to upload sensitive material merely to test whether the disabled capability might work. Any later rehearsal needs its own approved scope and safe test material.

Understand failures and suspension

A provider outage, expired evidence, incompatible profile, stale workspace policy or database-readiness failure can prevent use. Retrying cannot repair missing authority or make an incompatible setup valid.

The responsible support owner should identify the failing stage and preserve the original operation reference. Governance and operational owners decide containment and any later reapproval. Private failure must not silently send content to a public provider, another region or another boundary.

If an earlier request has an uncertain outcome, check that request's saved state and accounting evidence. See credits and uncertain outcomes. Do not create another paid request to avoid resolving the first one.

Plan recovery and rollback

Agree recovery objectives, the responsible owners and what evidence proves restoration before relying on the service. A restored database is not ready merely because it starts.

The recovery must preserve tenant isolation, current source access, withdrawn/deleted information, supplier validity and compatible search data. The implementation rejects activation of retired or tombstoned generations. Restoring an old copy must not resurrect content that is no longer permitted.

Returning to a previous qualified configuration still requires current approval and compatible data. Moving from a private boundary to a public service is a new deployment decision. It is not an automatic recovery option.

Plan retention and deletion

Identify all copies involved: original records, prepared search data, operation evidence, exports, backups and restored copies. Ask the retention owner how holds and deletion requests apply to each.

The private search library contains removal methods, but those methods do not by themselves decide whether deletion is authorized or complete. Operators must verify the approved outcome and guard against rebuilding removed information from an outdated source or backup.

Members should use the product's supported data-request and support processes. They should not be asked to change storage or infrastructure settings.

Prepare a support or decision handoff

Provide the workspace context, intended task, safe source or operation reference, approximate time and the status you observed. Say whether the question concerns disabled availability, an existing commercial review, an ongoing incident or a future assessment.

For a business decision, request a clear statement of supported scope, unresolved implementation work, accountable owners, current evidence, recovery commitments and the next decision. This documentation does not supply an activation date or a deployment approval.

For an existing operation, retain its identity until the authoritative result and accounting outcome are established. Support should explain the next step in product language and keep credentials, protected content and internal diagnostics in the appropriate restricted channel.

Zerohive documentationStaging draft · 22 Oct 2026

Search documentation

Search guides, platform concepts and tasks.

↑ ↓ to move Esc to close