Reviewer includes AI-powered features that help you extract, classify, and analyze regulatory content faster. This guide explains each AI feature, where to find it, and how to use it effectively.
Where AI Appears in Reviewer #
| Feature | Location | What It Does |
|---|---|---|
| AI Search Answers | Global Search | Synthesizes an answer to your question from document content |
| AI Enrichment | Reports (Chronology, Labeling, Correspondence, Specification, Clinical) | Extracts and classifies data from documents automatically |
| Content Crawl | Correspondence Tracking | Deep-reads document text to find meetings, commitments, and RFIs |
| Change Analysis | Labeling History Report | Compares document versions and highlights additions, deletions, and modifications |
| Health Score | Specification Dashboard | AI-calculated quality profile score based on Module 3 analysis |
AI Search Answers #
When you type a question in Global Search (e.g., “What are the storage conditions for the drug product?”), the AI analyzes your dossier documents and returns a synthesized answer.
The answer appears in an AI Answer panel above the search results, with:
- A plain-language answer synthesized from multiple documents
- Source references — clickable links to the specific documents used
- A confidence indicator
AI Enrichment in Reports #
Most reports have an AI Enrich button in the toolbar. Clicking it triggers the AI to analyze the dossier documents and populate the report with extracted data.
How AI Enrichment Works #
- Open any report (Chronology, Labeling History, Specification, Correspondence, or Clinical Studies).
- Select a dossier from the application selector.
- Click AI Enrich in the toolbar.
- The AI reads the relevant Module documents and extracts structured data (dates, types, names, values).
- Extracted data populates the report tables and charts. A progress indicator shows the enrichment status.
- Review the AI-extracted data. You can edit, correct, or delete any entry that the AI got wrong.
Correspondence Content Crawl #
In the Correspondence Tracking report, the Crawl Content button performs a deep AI analysis that goes beyond structural metadata. It reads the full text of Module 1 documents to identify:
- Meeting references and participants
- Questions from the agency (RFIs)
- Commitments and agreed-upon actions
- Response cross-references
This is more thorough than the Scan M1 step (which only looks at structural metadata) but takes longer to process.
Labeling Change Analysis #
In the Labeling History Report, AI Enrich compares different versions of the same labeling document and highlights:
- Additions — New text added in the later version
- Deletions — Text removed from the earlier version
- Modifications — Text that was changed between versions
This is especially useful for tracking labeling revisions across amendments and supplements, where manual comparison of 50+ page prescribing information documents would be extremely time-consuming.
Specification Health Score #
The Specification Dashboard’s Health Score is an AI-calculated metric that analyzes Module 3 (Quality) documents and scores the completeness and quality of the drug specification package. The AI extracts specification parameters, stability data, batch results, and impurity levels, then evaluates them against ICH guidelines.
The score is informational — it highlights areas that may need attention during review, not definitive quality judgments.
Best Practices for Using AI Features #
- Run AI Enrich once per report after importing a new submission. The AI needs document content to work — empty dossiers produce empty results.
- Always verify AI output. AI extraction is accurate for structured content (dates, numbers, form data) but may misclassify free-text content. Review and correct as needed.
- Use Refresh > Full Re-extraction if AI results seem stale or incomplete. This forces a fresh read of all documents.
- Manual entries override AI. If you manually correct an AI-generated entry, your correction is preserved on future enrichment runs.
