PDF Intelligence API

Visual Integrity PDF Intelligence API turns PDF drawings into structured, machine-readable intelligence — not just images. While vision models approximate document content from rendered pixels, this API reads the page's native vector structure directly: every object, coordinate, layer, and attribute, exactly as authored. Built on 30 years of conversion technology trusted by Oracle, Delta Air Lines, Amazon, BASF, and Siemens, it's the only Azure-native API that converts PDF files directly to vector formats and extracts all the data from them.

AI-Powered Document Analysis
An integrated analysis endpoint automatically characterizes every uploaded PDF — identifying document type (vector, scanned raster, hybrid, geo, document), quantifying object composition, detecting layers, evaluating text editability, and recommending the optimal conversion strategy. Once the object data is extracted, bring your own AI LLM — Claude, GPT, or Gemini — to get an instant analysis of the JSON content. Write your own prompts to query the drawing and act on it at the object level.

Object Extraction
The Object Extraction endpoint returns every vector object from a PDF page as structured JSON — polygons, polylines, text, and images with typed geometry, mm-precision coordinates, color values, and layer assignments. Purpose-built for engineering analysis, GIS data recovery, automated QA, and feeding structured drawing content directly into your own LLM pipelines or data systems.

File Conversion
Convert to 16 output formats with 46 configurable options:

• CAD: DWG, DXF, HPGL
• Vector: SVG, EMF, WMF, CGM, EPS
• Raster: PNG, JPG, TIFF, BMP, GIF
• Document: PDF, PostScript, Text

Format-specific options include AutoCAD version targeting, DPI control, layer assignment modes, CMYK output, compression selection, text-combining strategies, and shape recognition, all controllable on a per-request basis via the API.

Developer Experience

• Interactive demo with file conversion, object extraction visualization, and API explorer — try all endpoints directly at the API URL
• OpenAPI 3.0 documentation with complete schema definitions
• Simple REST workflow: upload → poll → download, with secure time-limited download links
• API key authentication with per-key credit tracking

Pricing
Credit-based, pay-per-page pricing with no minimums, no setup fees, and no monthly commitments. Each page converted or extracted consumes credits from the subscription balance. Ideal for evaluation at low volume and scalable for enterprise workloads.

One thing worth a quick look before you save: the MS AI-generated version you pasted drops Delta Airlines and BASF from the trusted-companies list (only naming Oracle, Amazon, Siemens) — worth confirming that wasn't a deliberate trim on your end before this replaces it, since the version above keeps all five names as originally listed.