Use Adobe Acrobat's AI to Extract and Clean Up Document Text

What This Does

Adobe Acrobat's AI Assistant can read scanned PDF documents and answer questions about their contents, letting you quickly extract dates, names, subjects, and key facts from lengthy scanned records without reading every page. Combined with Acrobat's built-in OCR, this turns image-only scans into searchable, queryable documents.

Before You Start

• Adobe Acrobat (Pro or Standard) is installed and open
• You have a scanned PDF document (either a typewritten record or a printed document, not handwritten)
• Your institution has an Acrobat license (often included in Adobe Creative Cloud)

Steps

1. Run OCR to make the scan text-searchable

Open your scanned PDF in Acrobat. Go to Tools → Scan & OCR → Recognize Text.

Select In This File and click the blue Recognize Text button.

Wait for processing (typically 10–30 seconds per page).

What you should see: The document looks the same visually, but you can now click on text to select it. The scan is now searchable.

Troubleshooting: If the recognized text looks garbled, the scan quality may be too low. Try rescanning at 300 DPI minimum, or use Enhance Scans (also in Scan & OCR menu) before running OCR.

2. Open the AI Assistant panel

Click the AI Assistant button in the right sidebar, or go to View → Tools → AI Assistant.

If this is your first time, you may need to sign in with your Adobe account and accept terms. The feature may require a paid Acrobat plan; check your subscription level.

What you should see: A chat panel opens on the right side of the screen.

3. Ask questions about the document

In the AI Assistant chat box, type questions about the document's contents. For archival work, useful queries include:

• "What dates are mentioned in this document?"
• "Who are the people mentioned in this document?"
• "Summarize the main topics covered in this document."
• "What is the subject of this letter?"
• "List any organizations mentioned."

What you should see: AI Assistant provides an answer based on the document's text, often with page citations.

4. Use the answers to inform your metadata

Copy the AI-extracted facts into your metadata record or finding aid notes. Review for accuracy against the actual document before entering into ArchivesSpace.

Real Example

Scenario: You're processing a box of 1940s–1960s typewritten board meeting minutes: 200 pages of scanned PDFs. You need to write series-level description noting key topics and date ranges.

What you do: Open one multi-year PDF in Acrobat, run OCR, then ask AI Assistant: "Summarize the main topics discussed in these meeting minutes and list the date range covered."

What you get: A summary of major topics (budget approvals, personnel decisions, building projects) with the date range identified. Enough to write a series-level description without reading all 200 pages.

Tips

• AI Assistant works best with typewritten or clearly printed documents; handwritten documents need Transkribus instead
• The AI cites pages for its answers. Click the citations to verify the source before using the information
• For very long documents, ask about specific sections rather than "summarize the whole document"

Tool interfaces change. If a button has moved, look for similar AI/magic/smart options in the same menu area.