Understanding Records information in HiveAI¶

Records can supply HiveAI with current record information and references to eligible documents. This supports answers grounded in the organization's records and a basic check of a record's current status. The records system remains responsible for the record, its documents and its lifecycle.

Availability: this page explains the implemented Records contribution to HiveAI. Connected assistant screens, organization-specific access and successful processing in a target environment still need verification. It does not provide unverified button-by-button instructions or promise that every record is searchable. Broader assistant actions and protected personnel journeys have separate rules and remain outside this explanation.

What information can be used¶

For an eligible record, the knowledge path can supply its subject and description together with references to documents that have a current version and the required file information. It uses current document versions; it does not offer the entire document-version history as one knowledge source. A record with incomplete document information may therefore have less usable document content than its file count suggests.

The basic status check is narrower. It returns the record's reference and current lifecycle, filing and custody information when access succeeds. It does not return the subject, description or document body through that status result. A status answer and an answer drawn from document contents are different operations.

An assistant reference is a way to trace the answer's source. Whether its link opens the intended application page still needs to be verified in the connected journey. Keep the official record reference when following up with the registry.

Which records are excluded¶

This Records knowledge path excludes personnel records and records marked Secret in its classification data. It also excludes records that are archived, rejected or have completed final disposition. Such a record can still exist in the records system even when the assistant cannot retrieve it.

Closing a record alone does not automatically exclude it. Neither does its origin as a pre-system record. Conversely, receiving or registering a record does not establish that its document contents have already been processed for the assistant.

These are rules for this knowledge path. They do not establish access to personnel records through another assistant journey, turn a filing category into a complete confidentiality policy, or prove that all older indexed content has already been removed. Administrators should verify the organization’s classification and access controls as part of acceptance.

Access depends on the person and the record¶

The basic status and knowledge-access checks require an access context for the same organization, the required Records viewing permission and an allowed classification level. The source also checks responsibility or scope: for example, the creator, registering person, intended recipient, responsible office or position. An organization-head role has broader scope within those preceding requirements.

Receiving an external submission receipt does not grant internal assistant access. Seeing one record through an authorized role does not establish permission to view all records, file them, change their destination or dispose of them.

If access is denied, the normal source result withholds the record label and link. Do not interpret an unavailable answer as confirmation that no record exists. Support must use an authorized Records journey to distinguish access, lifecycle, missing information and processing problems.

How updates reach the assistant¶

Records publishes information about relevant changes for background processing. It can later provide a current snapshot when the knowledge service requests one. Before allowing a result to be used, the source compares the requested record and access versions with its current information.

A detected mismatch causes that result to be withheld. When the knowledge service requests an old snapshot, Records also announces the current source for processing. An access check can reject an old version without starting that refresh. These checks protect against treating an identified old version as current; they do not guarantee that a replacement answer is available immediately. A successful save, a change notification, completed background processing and an answer shown to a reader are separate milestones.

Some record changes do not immediately announce a knowledge update. A recovery scan can help refresh the source information, but there is no verified universal delay promise in this documentation. When a recent change is missing, check the authoritative record and keep the time of the change for support.

What a completed recovery scan means¶

A background scan works through records in bounded groups. Its completion records how far the source scan progressed and how many source notifications it published. It does not certify that every document was read, indexed or made available in an answer.

The current scan can skip an individual item that exceeds its configured size estimate and still advance past that item. Records can also change while the scan runs. Administrators should reconcile omitted or stale records rather than accept the completion flag as a certificate of complete coverage.

The size estimate used by this scan is not the total size of the stored files, an AI usage statement or a price. This page establishes no charge, credit amount or processing-time commitment.

Read status in the right context¶

The basic status result can help distinguish a registered record from one in processing or already filed. It cannot establish that all work is finished, that a document has been physically handed over, or that a retention requirement has been met.

In particular, the source's simple disposition status can say “Eligible” when it merely found no completed disposition and no block. That check does not calculate the retention period or evaluate the current retention policy. A connected assistant must not turn that value into permission to archive, close or destroy a record.

Use classification, retention and final disposition for the actual eligibility and execution rules. The current archive/close execution does not delete stored document bytes. Records that have already completed disposition are excluded from this basic knowledge/status path, so their completion should be checked through the authorized records system.

When an answer is missing, old or unexpected¶

Check the record through your organization's authorized Records journey. Confirm the organization, record reference, recent changes and whether the record belongs to an excluded category or lifecycle state. If you cannot access that journey, ask the registry or support team to investigate within its permissions.

Give support the record reference, approximate time, action or question attempted and any safe error message. Do not send document contents, public tracking secrets or another person's personnel information through an unrelated support channel. Avoid changing classification or resubmitting a record just to make it appear in the assistant.

Support should distinguish a source-access refusal, stale information, missing document details, incomplete background processing and an application-link problem. A retry is useful only when it addresses the actual cause; repeated submissions do not prove that the knowledge source has been repaired.

What administrators and business owners should verify¶

Before relying on this capability, test representative allowed and excluded records with the intended roles. Check a recent document update, a changed responsibility, a withheld result and a missing source. Confirm how readers return to the authorized record and who investigates discrepancies.

Treat source freshness, processing completeness, access and the quality of an answer as separate acceptance questions. The underlying code and draft documentation alone do not establish an environment's delivery, an organization-wide knowledge inventory, a compliance certification or reader acceptance.