Deep Analysis
Learn how to use Agentive's Deep Analysis functionality to perform AI-driven reviews of audit documents.
Use Deep Analysis to chat with audit documents in essay-style format with AI-powered insights and side-by-side source citations.
Deep Analysis provides ChatGPT-style conversations specialized for audit standards with automatic citation linking to source documents.
Set up Deep Analysis
Create the request
- Navigate to New Request
- Enter a descriptive title (e.g., "Deep Analysis - BOD Minutes")
- Click Create
Configure document group
- In Getting Started, click Add Table or select No Table if no structured sample is needed
- Under Document Group, enter a descriptive name (e.g., "Board Minutes")
- Click Create
Deep Analysis works with or without a sample table—use it for unstructured document review.
Write effective prompts
Access Ask AI
- Locate the Ask AI button at the top-middle of the screen
- Click to open the chatbot interface
- The AI helps you craft audit-focused prompts
Generate the prompt
Describe your audit objective to the AI:
Example input:
I am an auditor reviewing board minutes in accordance with AICPA audit standards.
Can you write a prompt that helps me extract that information?The AI will generate a comprehensive prompt tailored to your audit area and standards.
Refine and copy
- Review the AI-generated prompt
- Provide feedback to include specific items (e.g., "flag unusual or fraudulent transactions")
- Highlight the finalized prompt text
- Use
Ctrl+C(Windows) orCmd+C(Mac) to copy
Use Ask AI to "wordsmith" complex prompts that hit all key audit compliance points.
Run Deep Analysis
Upload documents
- Go to the Request view
- Drag and drop relevant PDF files into Requested Files
- Wait for OCR processing to complete
Deep Analysis requires OCR-processed documents. Wait for the "Ready" status before proceeding.
Configure analysis scope
Single summary across all files
Best for:
- Board minutes across multiple quarters
- Comprehensive policy reviews
- Trend analysis across documents
- Executive summaries
How it works:
- AI reads all documents as a single corpus
- Generates one comprehensive analysis
- Identifies patterns across files
- Produces narrative summary
Use this mode for high-level reviews and pattern identification across multiple documents.
Transaction-level analysis
Best for:
- Vendor invoice reviews
- Individual contract analysis
- Transaction-specific questions
- Detailed line-item review
How it works:
- AI processes each document separately
- Creates individual chat logs per document
- Provides transaction-specific details
- Enables granular comparison
Individual analysis creates a separate chat summary for every document in the group.
Add the analysis
- Click Add Deep Analysis in the sidebar
- Choose your scope (all together or individually)
- Paste your prompt into the Deep Analysis Question text box
- Click Add Deep Analysis
Execute the run
- Click the Play button on the Deep Analysis entry
- Or click Run from the workflow view
- Wait approximately 60 seconds for AI processing
Processing time varies based on document count and complexity. Most analyses complete within 1-2 minutes.
Review results
Read the analysis
- The AI-generated summary appears in the chat window
- Review the narrative response for audit findings
- Look for flagged items based on your prompt
Navigate with citations
- Click numerical Citations within the text (e.g., [1], [2])
- The source document automatically opens to the exact referenced page
- Review the original evidence side-by-side with the analysis
Citation linking enables rapid verification—click to see exactly where the AI found the information.
Refine the output
If the response needs adjustment:
- Use the Ask anything... box to provide follow-up instructions
- Example: "Can you please make this more concise and summarize the findings in 4 paragraphs?"
- The AI regenerates the response based on your feedback
Copy for documentation
- Hover over the Copy button
- Click to copy the text
- Paste into external reports or Word documents
Analysis scope comparison
Use single summary for:
Board minutes
Review all quarterly minutes to identify significant decisions or commitments.
Policy documents
Analyze complete policy manual for compliance with audit standards.
Contract sets
Review multiple related contracts to assess overall terms and risks.
Trend analysis
Identify patterns across multiple months or quarters of similar documents.
Example prompt:
Review all board minutes for evidence of significant transactions,
related party dealings, or commitments that may impact financial statements.
Flag any unusual or potentially fraudulent activities.Use individual analysis for:
Vendor invoices
Assess each invoice for expense category, prepayments, or unusual items.
Employee contracts
Review individual employment agreements for compensation details.
Transaction-specific
Answer questions unique to each document in the set.
Detailed comparison
Compare specific elements across documents (terms, dates, amounts).
Example prompt:
What is the nature of this transaction? Does it appear to be a prepayment,
capital expense, or operational expense? Identify the service date range
and vendor name.Best practices
Be specific in prompts
Clearly state your audit objective and what you're looking for (fraud indicators, unusual transactions, compliance checks).
Reference audit standards
Mention relevant standards (AICPA, PCAOB, ISA) to guide AI analysis.
Start broad, then refine
Run initial analysis, then use follow-up questions to dig deeper into flagged items.
Verify citations
Always click citations to confirm AI findings against source documents.
Use for supplemental review
Deep Analysis complements but doesn't replace structured workpaper tables.
Choose appropriate scope
Use "all together" for summaries, "individually" for transaction detail.
When to use Deep Analysis
Perfect for:
- Unstructured documents: Board minutes, contracts, policies where structured data extraction isn't applicable
- Fraud detection: Flagging unusual transactions or patterns across documents
- Compliance reviews: Assessing adherence to audit standards or client policies
- Narrative summaries: Creating executive summaries of complex documents
- Follow-up questions: Digging deeper into findings from structured extraction
- Pattern identification: Finding trends across multiple similar documents
Deep Analysis excels at narrative review and pattern recognition beyond simple data extraction.
Use other tools for:
- Structured data extraction: Use AI prompts for specific field extraction
- Matching to samples: Use document matching for invoice-to-sample linking
- Calculations: Use formula columns for mathematical operations
- Specific field capture: Use lasso tool for manual extraction
- Repeatable workflows: Use AI prompts and formulas for consistent processing
Deep Analysis results don't populate workpaper tables automatically—use it as a supplemental review layer.
Related guides
AI Prompts
Extract structured data from documents using targeted prompts.
Document Matching
Automatically match documents to sample rows before analysis.
Lasso Tool
Manually extract specific values when automation needs precision.
Create Requests
Set up requests that can include Deep Analysis questions.
How is this guide?