Skip to main content

Document Intelligence Agents

Document Intelligence is a pair of pre-built Marketplace agents that transform raw, unstructured documents into structured, benchmarked, and insight-rich outputs. They are designed for enterprise document sets such as annual reports, ESG filings, contracts, compliance documents, and operational reports — files that are long, inconsistently formatted, and spread their data across text, tables, charts, footnotes, and annexures.

The two agents are built to work together:

  1. Data Extraction Agent — extracts accurate, structured data from any type of PDF.
  2. Data Analysis & Insights Agent — analyzes the extracted data to generate comparisons, benchmarks, and executive-level insights.

Used as a pipeline, they remove the manual effort of reading, validating, and consolidating documents, so teams can move straight to decision-making.

Who it's for​

The agents support any function that works with high volumes of complex documents, including Finance, ESG, Compliance, Procurement, and Strategy. Typical jobs include benchmarking companies against one another, tracking metrics across reporting periods, and producing consolidated packs for leadership and board reviews.

Data Extraction Agent​

This agent understands and extracts information from documents.

What it does:

  • Accepts one or more PDFs — annual reports, ESG reports, contracts, or policy documents.
  • Identifies relevant data even when document formats, terminology, and labels differ.
  • Extracts key metrics from text, tables, charts, and structured sections.
  • Produces clean, standardized output with clear references back to where each value came from.

Inputs: one or more PDF documents.

Output: a consistent, reliable dataset of extracted metrics with source traceability — regardless of how the original documents were structured.

Data Analysis & Insights Agent​

This agent operates on the extracted data and turns it into intelligence.

What it does:

  • Compares values across multiple documents or companies.
  • Highlights differences, trends, and performance gaps.
  • Provides internal comparisons alongside external benchmark context.
  • Generates explanations, risk indicators, and recommendations.
  • Produces consolidated summaries suitable for leadership review.

Input: the structured dataset produced by the Data Extraction Agent.

Output: interpretation and direction — comparisons, benchmarks, insights, and executive-ready summaries, not just raw numbers.

End-to-end flow​

  1. Upload one or more PDF documents.
  2. The Data Extraction Agent identifies and extracts the key information.
  3. Extracted data is organized into a consistent structure.
  4. The Data Analysis & Insights Agent analyzes that data.
  5. The system produces comparisons, benchmarks, insights, and summaries.
  6. You receive a complete intelligence pack, ready for decision-making.

Why use these agents​

  • Structured, reliable data extracted from any type of PDF.
  • Consistency across documents, even when formats differ.
  • Cross-document comparison without manual consolidation.
  • Benchmarking context to understand performance and position.
  • Executive-ready insights instead of raw numbers.
  • Confidence and traceability — extracted values can be traced to their source.
  • Scalability — process many documents efficiently.

Using and cloning the agents​

The Data Extraction Agent and Data Analysis & Insights Agent are published as pre-built agents in the Marketplace. To use them in your own project, find each listing in the Marketplace tab, open a sandboxed trial session to try it, then clone it into your chosen project. See Cloning an agent for the full flow, including dependency resolution and credential setup.

Once cloned, each agent runs like any other Application you've built and can be customized to your document types and metrics.