Agentic Table Extractor
Enterprise Feature
This feature is available in Unstract Cloud Edition and Unstract On-Premise Edition.
Agentic Table Extractor is a multi-agent, VLM-powered pipeline within Prompt Studio that extracts structured table data from PDFs, Excel files, and images. It uses a 6-agent pipeline with two LLM adapters (advanced + lite) and an X2Text adapter to deliver accurate table extraction, even from complex multi-page documents.
Documentation
- Introduction. What Agentic Table Extractor is, how it differs from standard table extraction, and the 6-agent pipeline overview.
- Getting Started. Prerequisites, step-by-step setup, and running your first extraction.
- Writing Prompts. Prompt structure, the readiness checklist, and tips for effective prompts.
- Configuring Settings. Lite LLM adapter, page range, and parallelism settings.
- Excel Support. Working with
.xlsxand.xlsfiles, including multi-sheet handling. - JSON Schema Transformation. Defining output schemas and how code generation transforms raw output.
- Best Practices & Troubleshooting. Optimization tips, known limitations, and common issues.