Note: Veeva did not create the LLM that Quick Check Agent uses, nor do we train it using your data. Our implementation is designed to work for all customers.
to open the Quick Check Agent panel.
Note: The Quick Check Agent quality check outcomes are not audited, and are not reportable. Quality check results are specific to the document version being assessed.
### Performing Quality Checks
When you open the panel, Quick Check Agent immediately begins performing the following predefined quality checks on the document:
Spelling
: This check identifies potential spelling errors in the document, focusing on misspellings, common misuses, and typos, while considering industry-specific terminology.
Grammar
: This check identifies potential grammatical errors in the document, such as verb tense, punctuation, pronoun and subject-verb agreement, sentence structure, and modifier placement.
Phrase Assessment
: This check identifies wording and phrases that may require further scrutiny, including those suggesting promotional or potentially misleading language. Quick Check Agent integrates directly with your _Claims Library_, automatically treating _Eligible Claims_ as _Allowed_ to prevent false positives during this check.
Boxed Warning (United States Only)
: This check determines whether the document's _Product_ requires a Boxed Warning according to the FDA database of Boxed Warning products.
: If Quick Check Agent determines that a Boxed Warning is required, it first detects if one is present, and, if so, evaluates its format and location.
: Quick Check Agent does not review the text of the Boxed Warning.
: If the document is not for the United States, the Boxed Warning check is hidden and does not run.
Note: If your Product is not yet listed in the FDA’s database (typical for premarket products), you can mark your Product as an anticipated Boxed Warning product.
Note: Veeva defines these quality checks. You cannot edit these checks or add custom ones.
To clear Quick Check results and run them again, click **Re-run the Quick Check Agent**.
### Tailoring Quick Check Agent with Learned Terms
To improve the quality of results relating to your organization, Quick Check Agent can be tailored by providing it with your organization's specific terminology and phrases as additional context. When reviewing detected spelling errors, grammar errors, or phrases for review, users can convert specific findings into _Allowed Terms_, effectively telling Quick Check Agent not to raise these items again in the future. By default, the _Allowed Term_ record is tagged with the document's _Brands_, _Countries_, and _Language_ (on multi-lingual Vaults), along with the specific quick check it was added from, meaning the _Allowed Term_ only applies to documents matching those attributes. Business Admins can modify these attributes.
In addition, from the _Text Asset_ object where _Allowed Terms_ are managed, Business Admins can also define _Prohibited Terms_. This tells Quick Check Agent to always flag these items if they are found in matching documents (even if the AI normally would not catch them).
When you select an issue in the Quick Check Agent, the Issue menu opens to display the following actions:
* **Copy Suggestion**: Copies the suggested text to the clipboard.
* **Annotate**: Creates an annotation on a document. Users can annotate all Quick Check issues across all categories.
* **Learn Term**: Creates a new _Allowed Term_ record based on the selected text, automatically populating the record with all the document's Brand and Country values.
Alternatively, with the menu open, you can quickly execute an action using keyboard shortcuts by pressing **C** for _Copy Suggestion_, **A** for _Annotate_, and **L** for _Learn Term_.
#### Dynamic Instructions
_Allowed Terms_ and _Prohibited Terms_ are programmatically identified after the quick checks have produced their results, augmenting the final result set. However, Business Admins can also choose to enable a Yes/No field on specific _Allowed Terms_ and _Prohibited Terms_ to include them in the _Phrase Assessment_'s detection logic. By doing so, the _Phrase Assessment_ check looks for terms like these during its initial run by looking for semantic matches and flagging them accordingly.
### Matching Websites {#websites}
When performing the _Privacy Policy Link_ and _Unsubscribe Link_ quality checks, Quick Check Agent attempts to match any extracted links to existing _Website_ object records based on the _Country_ and _Language_ fields on the document and object records. If the _Country_ and _Language_ fields are blank on a _Website_ record, Quick Check Agent considers the record to apply to all values.
## Using the Content Agent
The Content Agent that powers the Veeva AI Chat is Veeva-aware, meaning that it has a deep structural knowledge of Vault and its functionality. The agent reviews all document fields for expanded context and distinguishes between a document's content and its document fields. This allows the agent to provide accurate responses quickly. The Content Agent also does not guess when reviewing ambiguous data in charts or images. If a value is unclear, the agent reports this to the user and sticks to objective analysis. The agent's responses are grounded in the document's content and its own Vault-based knowledge.
### Using the Veeva AI Chat
Veeva AI for PromoMats also provides access to an intuitive AI Chat, powered by the PromoMats Assistant, which you can access by clicking 
in the upper right of the Vault page. You can ask the Assistant questions about the specific document version you are viewing. Responses are generated from the document's text, images, Quick Check Agent results.
To use the AI Chat, enter your message in the text box at the bottom of the chat. Press **Send** (
) or press the **Enter** key to send your message, or press **Shift** + **Enter** for a new line. You can enter a maximum of 5,000 characters per query. The Assistant displays 
while generating your response.
The Veeva AI Chat automatically determines which chat-enabled Agent action, _Ask Questions_ or _Analyze Images_, should be used to respond to your question or request.
### Using the Reviewer Summary
The _Reviewer Summary_ for the Content Agent includes automated discrepancy detection to cross-reference document text with metadata, which flags mismatches in brand, market, and audience intent. This can highlight critical errors for such things like mislabeled documents.
The _Complexity Assessment_ section can calculate a review difficulty rating based on the density of the clinical data and visualizations, which can help reviewers estimate the effort required.
The _Key Fields_ section offers a side-by-side view of the system metadata and AI-analyzed content with status indicators to help accelerate verification. The _Key Fields_ section also can include the _Document Metadata Fields_ context item.
## Related Permissions
The following [permissions](/en/lr/22824/) control your ability to use Quick Check Agent:
| Type | Permission Label | Controls |
| --- | --- | --- |
| Document Role | View Content | Ability to view and use the **Quick Check Agent** panel on a document version in a given lifecycle state |
| Document Role | Annotate | Ability to save an identified issue as a document annotation |
[1]: #websites