Supported Formats¶

Complete reference of file types you can upload to your knowledge bases.

Overview¶

Functional AI supports 21 file formats across documents, code, and data. All files are processed to extract text content for semantic search.

Key principle: If it contains extractable text, it can be used.

Images Not Processed

Images, charts, and diagrams within documents are not processed. Only text content is extracted and indexed.

Document Formats¶

PDF Files (.pdf)¶

Aspect	Details
Best for	Manuals, reports, documentation, whitepapers
Text extraction	Excellent for text-based PDFs, poor for scanned images
Max size	500 MB

PDF Quality

Use PDFs with selectable text rather than scanned images. If you have scanned documents, run OCR first.

PDF Optimization & Issues

Optimization tips: - Remove cover pages and blank pages - Compress images to reduce file size - Convert image-heavy PDFs to text-only versions - For scanned documents, use OCR tools - Test if text is selectable by trying to highlight it

Common issues: - Scanned PDFs appear as images - no text extracted - Password-protected PDFs cannot be processed - Very large PDFs (100+ MB) take significant time to process

Microsoft Word (.doc, .docx)¶

Aspect	Details
Best for	Business documents, policies, guides, procedures
Text extraction	Excellent
Max size	500 MB

Word Optimization Tips

Use clear headings (Heading 1, Heading 2, etc.)
Avoid text boxes and complex layouts
Remove unnecessary images
Use tables for structured data (converted to text)
Save as .docx (more reliable than old .doc format)

Plain Text (.txt)¶

Aspect	Details
Best for	Simple content, logs, notes, transcripts
Text extraction	Perfect (already plain text)
Max size	500 MB

Markdown (.md)¶

Aspect	Details
Best for	Technical documentation, README files, knowledge bases
Text extraction	Excellent (headings preserved as text)
Max size	500 MB

Why Markdown Works Great

Clean structure with headings makes chunking more effective. No complex formatting to interfere with text extraction.

Presentation Formats¶

PowerPoint (.pptx)¶

Aspect	Details
Best for	Slide content, training materials, presentations
Text extraction	Good (text from slides)
Max size	500 MB

PowerPoint: What Gets Extracted

Extracted: - Slide titles and body text - Bullet points and lists - Table content

Not extracted: - Images and diagrams - Charts and graphs - Embedded videos - Animations

Optimization tips: - Ensure critical information is in text, not images - Use slide titles effectively (they help with chunking) - Consider exporting to PDF for more control

Data Formats¶

JSON (.json)¶

Aspect	Details
Best for	Structured data, product catalogs, configurations, API responses
Text extraction	Excellent (preserves structure as text)
Max size	500 MB

Best use cases: - Product catalogs with descriptions - FAQ data in structured format - Configuration documentation

JSON Optimization

Tips for better results: - Keep nesting reasonable (3-4 levels max) - Use descriptive key names - Include text descriptions alongside data values

Example - Good JSON structure:

{
  "products": [
    {
      "id": "SKU-001",
      "name": "Premium Widget",
      "description": "High-quality widget with advanced features",
      "features": [
        "Durable construction",
        "2-year warranty"
      ],
      "faq": [
        {
          "question": "Is this dishwasher safe?",
          "answer": "Yes, top rack only"
        }
      ]
    }
  ]
}

Code Files¶

Code files are useful for technical documentation, API references, or coding assistants.

Language	Extensions	Best For
Python	`.py`	Python modules, API implementations
JavaScript/TypeScript	`.js`, `.ts`	Frontend code, Node.js modules
HTML	`.html`	Web pages, email templates
CSS	`.css`	Style documentation, design systems
Java	`.java`	Java API documentation
C/C++	`.c`, `.cpp`	Systems programming reference
C#	`.cs`	.NET application code
Ruby	`.rb`	Rails applications
PHP	`.php`	PHP application code
Go	`.go`	Go service documentation
Shell	`.sh`	Bash scripts, automation docs

All code files: Max size 500 MB, perfect text extraction

Code File Best Practices

Include comprehensive comments and docstrings
Remove credentials and secrets
Focus on well-documented, exemplary code
Consider if Markdown documentation might be clearer

Other Formats¶

LaTeX (.tex)¶

Best for academic papers and technical documents. Max size: 500 MB.

Complete File Type List¶

Category	Supported Extensions
Documents	`.pdf`, `.doc`, `.docx`, `.txt`, `.md`, `.pptx`
Data	`.json`
Code	`.py`, `.js`, `.ts`, `.html`, `.css`
Code (additional)	`.java`, `.c`, `.cpp`, `.cs`, `.rb`, `.php`, `.go`, `.sh`
Other	`.tex`

Format Recommendations¶

Best Formats for RAG¶

Use Case	Recommended Format
Product documentation	PDF, Markdown
FAQs	Markdown, Text
Policies	PDF, Word
Technical docs	Markdown
Data/catalogs	JSON
Code reference	Native code files

Formats to Avoid¶

Format	Issue	Alternative
Scanned PDFs	No text extraction	Use OCR first
Image files	Not processed	Convert to text
Video/Audio	Not supported	Provide transcripts
Spreadsheets (CSV, XLSX)	Not supported	Convert to JSON
Password-protected	Cannot read	Remove protection
Very large files	Slow processing	Split into sections

File Size Optimization¶

PDF Optimization

Tools: Adobe Acrobat, SmallPDF, iLovePDF

Techniques: 1. Compress images: Reduce image quality/resolution 2. Remove pages: Delete cover, blank, and unnecessary pages 3. Remove metadata: Author info, comments, etc. 4. Convert to grayscale: If color isn't essential

Example: A 50 MB PDF manual can often be reduced to 5-10 MB without losing text quality.

When to Split Files

Instead of one massive file, consider splitting when:

Single file exceeds 50 MB (easier to manage)
Content covers distinct topics
Updates affect only portions

Example: Instead of "Complete Product Manual.pdf" (100 MB), split into: - "Installation Guide.pdf" (5 MB) - "User Manual.pdf" (20 MB) - "Troubleshooting.pdf" (10 MB) - "Technical Specifications.pdf" (8 MB)

Next Steps¶

Upload Files to Your Store