Smart AI Assisted Workflow for PDF Import

 Jul 15, 2026 Lukáš Cerman news share

Do you need to import requirements from PDF files while preserving complex formatting, images, equations, and tables? We offer custom services and tools powered by the latest AI models to help you with this task.

Import Stakeholder Requirements From PDF

The PDF remains one of the most common formats for sharing customer requirements and standards because of its portability. However, importing PDF documents into a requirements management tool can be challenging when they contain complex formatting or are available only as scans.

In this post, we introduce our custom solution for importing PDF files into ReqView. It is powered by AI models to achieve the highest possible conversion quality and includes custom tools that allow users to conveniently review and control the conversion results.

Challenges With Import From PDF

The usual approach is to convert the PDF to DOCX and then edit it in MS Word to prepare it for import. The title page, page headers and footers, table of contents, and other unnecessary content should be removed.

This works quite well for PDFs that contain a selectable text layer and have a simple layout. However, it can be challenging in the following cases:

• Complex Layout: The PDF contains positioned text, lines, images, and drawing instructions. It does not reliably expose the document hierarchy, list levels, table structure, or requirement attributes needed after import.

  For requirements documents, any subtle conversion error can be potentially risky: a section can lose its parent heading, a page break can split a nested list, a table can become flat text, an equation can be misread, or a figure can be left behind. Moreover, requirement attributes, such as rationale, can be attached to the wrong requirement object if the structure is inferred incorrectly.

  Some PDFs also contain incomplete text layers. For example, a visually obvious heading or section number may be stored as an image instead of selectable text and may be missing from the imported document.

• Scanned PDFs: The PDF file was scanned from paper and stores each page as a bitmap. You need to use an Optical Character Recognition (OCR) tool to convert the bitmaps into text and other structured information. The latest versions of MS Word and Adobe Acrobat have built-in OCR functionality. You can also use online OCR tools, such as Google Docs or Adobe Acrobat PDF to Word Converter. However, the correct interpretation of the complex document layout remains challenging.

Our goal is not just to extract the text from a PDF but to verify that the resulting structured requirements document matches the PDF content before it becomes a managed set of requirements in ReqView.

Our AI Assisted PDF Import Solution

To minimize the manual effort required to detect and fix the PDF conversion errors, we have developed a custom conversion tool powered by the latest AI models to achieve the highest possible conversion quality. In addition, we provide an easy-to-use review tool that enables users to inspect and correct the conversion results before importing them into ReqView.

Our AI-assisted solution consists of several conversion steps and uses HTML as an intermediate format between the original PDF document and the resulting ReqView project. We have implemented a dedicated tool for each step. This approach makes the conversion process transparent and gives users full control over the final output.

1. Convert PDF Document to Intermediate HTML Document:

In the first step, the user runs our custom Converter tool, which uses a Large Language Model (LLM) provided by OpenAI or Anthropic to convert the source PDF document into an intermediate HTML document. To ensure maximum security, customers manage the access to the AI providers and retain control over data usage, including ensuring that their data is not used by the providers to train AI models.

The intermediate HTML document uses a small subset of HTML tags to preserve the document structure and content required for import into ReqView. This includes section headings, requirement statements with rich-text formatting, lists, tables, images, and equations.

We offer customization of the Converter tool to meet specific customer needs for a specific document layout.

2. Review Intermediate HTML Document:

To keep human control in the process, the user should verify the correctness and completeness of the conversion performed by AI in the previous step. In this step, the user uses the Import Reviewer tool to quickly identify and correct any conversion errors before importing the document into ReqView.

The Import Reviewer offers several easy-to-use features that minimize the user’s effort during the review:

• Side by Side View: The Import Reviewer displays the source PDF document and the intermediate HTML document in panes with synchronized scrolling. This gives the user the full context needed to review the imported text and document structure, including headings, nested lists, tables, equations, and images.

  

• Highlighting of Differences: The Import Reviewer highlights differences between the PDF document and the intermediate HTML document. This allows the user to focus on correcting conversion errors.

  The screenshot below illustrates the highlighting of changed or extra fragments directly in the comparison view, which is very useful for fixing OCR-like mistakes, missing words, and misplaced fragments.

  

• Editable HTML Source: The Import Reviewer displays the editable source of the intermediate HTML document and provides quick edit actions for transferring differences with a single mouse click. This allows the user to fix identified conversion errors and visually verify the result immediately.

The user can optionally store the intermediate HTML document in a version control system, such as Git or Subversion (SVN), to maintain a complete history of manual corrections. This makes it easy to review changes, trace individual edits, or revert to an earlier version if necessary.

3. Convert Intermediate HTML Document Into ReqView Document:

After completing the review, the user runs the Importer tool to import the intermediate HTML document into the ReqView JSON data format. The tool extracts document sections and paragraphs from the HTML and converts them into ReqView document objects.

We offer customization of the Importer to meet specific customer needs for the specific document layout so that the document structure is correctly mapped to the corresponding ReqView object attributes such as original section number, rationale, verification description, etc., as demonstrated by the example in the next section.

Demonstration of NASA PDF Document Import

Let us demonstrate our AI-assisted PDF import solution using a public PDF document as an example. We chose the NASA document SSP 51721: ISS Safety Requirements Document because it provides a useful stress test: it is more than 300 pages long, structurally dense, and contains several types of information that need to be handled differently.

The Converter processes the hierarchy of sections and paragraphs, along with the requirement statements, rationales, and verification sections while ignoring the title page, page headers and footers, table revision history, preface, list of approvals, and the table of contents. The tool also preserves rich-text formatting, including nested lists, tables that span multiple pages, equations, and figures.

The example PDF document also illustrates how the Converter and Import Reviewer address common PDF conversion challenges, such as:

• Text Stored as Bitmaps: The example PDF document stores section numbers as text fragments in the PDF text layer only for the first two section levels. The section numbers at level three or lower are stored as bitmaps. The Converter needs to correctly interpret these bitmaps as numbers that are part of the section headings.

• Tables and Equations: The PDF stores tables and equations as text fragments, lines, and drawing operations rather than semantic structures. The Converter needs to reconstruct table rows, columns, cell content, and mathematical notation. The Import Reviewer highlights equations in the view to draw the user’s attention to areas that should be checked visually.

  For instance, the following screenshot shows how the Converter imported the table “4.14-2 AVERAGE FRONTAL AREA” from the example PDF while preserving its mathematical equations.

  

• Figures: The example PDF document contains figures that provide important context that cannot be recovered from text alone. The Converter automatically extracts figures from the PDF, while the Import Reviewer displays them alongside the converted content so that users can verify they have been extracted correctly and associated with the appropriate requirements. During the conversion into ReqView format, the Importer imports the figures as ReqView attachments to the corresponding requirement objects.

  

• Deeply Nested Lists: The example PDF document contains requirement descriptions with deeply nested lists that may span multiple pages, so the Converter needs to infer the correct list structure.

  For instance, the following screenshot shows how the Converter imported the PDF document section “4.3.6.1.1. VERIFICATION – ELECTROLYTIC CAPACITORS”, which contains four levels of nested numbered lists spanning two pages.

  

• Text Alignment: The example PDF document contains requirement descriptions that combine rich-text formatting, tables, and notes, so the Importer needs to align this information with the corresponding requirement text.

  For instance, the following screenshot shows how the Importer converted the PDF document section “4.3.1.2 WIRE DERATING” into a structured requirement in ReqView. It preserved the requirement text, the Table “4.3.1.2-1 WIRE SIZE DERATING AND CIRCUIT PROTECTION”, and the notes attached to the requirement. It imported the section ID, rationale, and verification information into the corresponding ReqView custom attributes.

  

Download: You can download the ReqView project created by our conversion tools and check for yourself how the imported document preserves:

• document hierarchy,
• requirement statements with rich-text formatting including nested lists, tables and equations,
• figures as image attachments to requirement objects,
• custom attribute Original ID,
• custom attribute Rationale, and
• custom attribute Verification.

Conclusion

We offer custom services to help you import PDF documents into ReqView including customization of the conversion toolset. Our solution uses the latest AI models to automate and accelerate the conversion from PDF. However, importing requirements still requires engineering review to ensure the converted document is complete, accurate, and structurally correct.

To support this process, we provide a custom review tool that allows engineers to compare the converted document with the original PDF, identify and correct conversion errors, validate the document structure, and only then import the reviewed document into ReqView.

Once the PDF document has been imported into a ReqView project, you can store the project in a Git or Subversion (SVN) repository, organize the imported requirements, and create links between requirements and other project artifacts. You can also export requirements from ReqView to MS Word, MS Excel, HTML, Jira Cloud, Azure DevOps, or ReqIF to share them with other stakeholders.