The Image
Intelligence
Layer for AI Agents
Viscribe extracts structured data from images using AI models. Define the output schema, pass the image, pick the model, and get parsed results back instead of free-form text.
Simple Image Calls, Structured Results
Install the library, point it at an image, and keep the result shape predictable.
1pip install viscribe1from pydantic import BaseModel, Field2from viscribe.images import extract3 4class Receipt(BaseModel):5 merchant_name: str | None = Field(description="Store or business name")6 total_amount: float | None = Field(description="Final receipt total")7 date: str | None = Field(description="Receipt date if visible")8 line_items: list[str] = Field(description="Visible purchased items")9 10result = extract(11 image_path="examples/venice.png",12 output_schema=Receipt,13 instruction="Extract the receipt fields visible in the image.",14 model_config={15 "model": "gpt-5-mini",16 "api_key": "sk-...",17 "temperature": 1,18 },19)20 21print(result.data.model_dump())Use Your Existing AI Model Providers
Keep your provider choices. Viscribe focuses the image workflow while you route through OpenAI, Anthropic, Mistral, Groq, OpenRouter, Fireworks, or your own compatible setup.
Vision Tools Built for Jobs
Five focused image methods designed around what applications and agents actually need to do.
Image methods
Image Description
Generate concise descriptions and tags from product images, screenshots, documents, or visual datasets.
Input
Output
A Venice canal scene with historic buildings, gondolas, and bright water framed by classic Italian architecture.
Vision Intelligence for Real Workflows
From product catalog automation to quality checks, Viscribe turns visual inputs into answers your application can use.
Product Catalog Automation
Extract product descriptions, attributes, categories, and variant details with a single schema-driven request.
Visual Monitoring
Track screenshots, public product images, and visual datasets for meaningful changes over time.
Agent Visual Reasoning
Give agents image tools that return focused answers instead of free-form prompts.
Quality Verification
Extract pass/fail checks, differences, and review notes from QA images using explicit schemas.
Document Data Extraction
Extract structured fields from receipts, forms, invoices, labels, and photographed documents.
Developer Workflows
Use one matching extract method across Python and TypeScript with clean result wrappers.
Free to Use, Open to Shape
Use Viscribe in your project, bring your own model provider, and share what would make it better for your workflow.
Open Source
Build image workflows with focused SDKs and your existing model provider.
Looking for a hosted solution?
Tell us what you want to build.