Skip to main content
Reducto Studio is a browser-based interface for building document processing pipelines. You configure parsing, extraction, and editing visually, test on real documents, then deploy to get a Pipeline ID you can call from code. This guide walks you through parsing your first document and deploying it for production use.

​
1. Open Studio

Visit studio.reducto.ai. You can explore the demo pipelines immediately without creating an account. To upload your own documents and create custom pipelines, you’ll need to sign up or log in.
If you’re part of an organization, your administrator must enable Studio access for your account.

​
2. Parse your first document

The goal here is to upload a document and see the structured output Reducto produces. Parse pipelines extract text, tables, figures, and document structure, returning everything as markdown with bounding box coordinates.
1

Create a Parse pipeline

Click Create pipeline in the top right, select Parse as the pipeline type, and give it a descriptive name.
Create pipeline dialog
2

Upload a document and run

Reducto supports 30+ file formats including PDFs, images, Word documents, spreadsheets, and presentations.
Upload document
Click Run to view the result of the document parse.
3

Review the results

The left panel shows your original document with colored bounding boxes overlaid on detected content. The right panel shows the parsed output as structured markdown.
Parsed results
The bounding box colors indicate content types: orange for text blocks, green for tables, purple for figures, red for headers, blue for key-value pairs, and grey for footers. This visual feedback helps you quickly assess whether Reducto correctly identified the document structure.

​
3. Adjust configurations (optional)

The default settings handle most documents well, but you can fine-tune parsing behavior in the Configurations tab. Studio offers two configuration modes: Simple mode provides quick toggles for the most common settings:
Simple configuration mode
  • Contains Handwritten Text: Enables agentic processing for better handwriting recognition
  • Enable AI Summarization: Generates descriptions of figures and charts
  • Return Figure/Table Images: Includes extracted images as URLs in the output
Advanced mode exposes the full configuration surface, organized into the same groups as the API:
Advanced configuration mode
GroupPurposeDocumentation
EnhanceAI-powered accuracy improvementsAgentic modes for text, table, and figure enhancement; Chart extraction for data from charts
RetrievalRAG and search optimizationChunking methods for segmenting content
FormattingOutput structure and stylingTable formats for markdown, HTML, JSON output; Additional document data for metadata
SpreadsheetExcel and CSV handlingSpreadsheet processing for multi-sheet documents
SettingsCore processing controlsOCR settings for extraction mode and language; Page ranges for partial processing
These options map directly to the API configuration parameters. See Parse Configurations for the complete reference. Each change takes effect when you click Run again, letting you iterate until the output matches your requirements.

​
4. Deploy for API access

Click Deploy when you’re ready to use your configuration from code.
Deploy dialog

Deploying a pipeline in Reducto Studio

Pipeline gives you a single Pipeline ID that encapsulates your entire workflow. Whether you built a simple parse or a multi-step flow (parse → extract → edit), your code just passes the ID and Reducto handles the rest. Update settings in Studio, redeploy, and all API calls automatically use the new configuration. Direct API Call exports the raw configuration as code. For multi-step workflows, this means separate API calls for each step, each with its own configuration block. Use this if you need to modify settings dynamically at runtime or want configs in version control. For most cases, use Pipeline—it’s cleaner code and lets non-engineers iterate on settings without redeploying your application. See Pipeline Basics for multi-step workflows.

​
Next steps

Parse Pipeline

Detailed configuration options for OCR, chunking, table extraction, and agentic enhancement modes.

Extract Pipeline

Define JSON schemas to pull structured data from documents, with automatic field detection and confidence scoring.

Edit Pipeline

Automate form filling using natural language instructions or structured field mapping.

Deploy to Production

Publishing workflows, API integration patterns, and monitoring deployed pipelines.