AI/LLM Capabilities

elDoc provides integrated AI and Large Language Model (LLM) capabilities across document management, document processing, enterprise search, knowledge retrieval, and workflow automation.

AI functionality is embedded directly into the elDoc platform and can be used to understand document content, retrieve relevant information, extract structured data, interact with enterprise knowledge using natural language, and perform document-related tasks through AI Agents.

The exact set of available AI capabilities depends on the elDoc configuration and the AI functionality enabled by the system administrator.

GenAI Chat

elDoc provides GenAI Chat for natural-language interaction with enterprise documents and information.

Users can ask questions, request summaries, analyze content, compare documents, retrieve specific facts, and interact with large document repositories without manually opening and reviewing every file.

GenAI Chat can use enterprise search and Retrieval-Augmented Generation (RAG) to provide responses based on information stored in elDoc.

Access to document content remains subject to the user's permissions.

Retrieval-Augmented Generation (RAG)

elDoc uses Retrieval-Augmented Generation (RAG) to combine Large Language Models with information retrieved from enterprise documents.

Instead of relying only on the general knowledge of the selected LLM, elDoc retrieves relevant document content and provides it to the model as additional context.

This allows generated responses to be grounded in information available in the organization's repository.

Agentic RAG

For more complex requests and large document collections, elDoc supports Agentic RAG.

Agentic RAG extends traditional RAG by allowing AI Agents to determine how information should be searched, retrieved, refined, and processed.

Depending on the request, Agentic RAG can:

  • perform multiple retrieval steps;

  • refine search queries;

  • combine information from multiple documents;

  • use keyword and semantic retrieval;

  • rerank retrieved results;

  • perform iterative reasoning;

  • use available tools before producing the final response.

This is particularly useful when the answer depends on information distributed across a large number of documents.

AI Agents

elDoc supports AI Agents that can perform document-related tasks based on natural-language instructions.

Depending on the available functionality and user permissions, AI Agents can perform tasks such as:

  • classify documents;

  • sort and group files;

  • rename files;

  • organize files into folders;

  • update document indexing values;

  • search for relevant information;

  • work with document content;

  • edit supported office files.

AI Agents combine LLM reasoning with controlled actions available within elDoc.

AI-Powered Document Indexing

elDoc supports AI Indexing for automatic extraction of structured document values.

When a document is associated with an Indexing Profile, AI Indexing can analyze its content and populate the corresponding Indexing Fields.

For example, a Bank Statement profile can contain fields such as:

  • Bank;

  • Account number;

  • Date;

  • Balance;

  • Currency;

  • Transaction Type.

AI Indexing can extract these values automatically and populate the structured indexing fields.

Extracted values can be reviewed, modified, and validated by users.

AI Document Processing

AI capabilities are integrated into the AI Document Processing module.

AI Document Processing, also known as Intelligent Document Processing (IDP), can use AI and LLM models for:

  • document classification;

  • structured data extraction;

  • form processing;

  • document understanding;

  • OCR;

  • recognition of complex and semi-structured documents.

This allows scanned and digitally generated documents to be transformed into structured, machine-readable information.

AI/LLM-Based OCR

elDoc can use AI and vision-capable LLMs for document OCR and image understanding.

This allows image-based files and scanned documents to be converted into machine-readable content that can subsequently participate in:

  • full-text search;

  • semantic search;

  • Document Indexing;

  • GenAI Chat;

  • RAG;

  • Agentic RAG;

  • AI Document Processing.

Semantic Search

elDoc supports semantic search, which retrieves information based on meaning and contextual similarity rather than only exact keywords.

For example, a query for:

employee compensation

can retrieve relevant documents discussing:

staff remuneration

even when the exact words used in the query do not appear in the document.

Semantic search complements traditional full-text search.

Hybrid Retrieval

elDoc can combine different retrieval methods, including:

  • keyword search;

  • full-text search;

  • vector-based semantic retrieval;

  • reranking.

This hybrid approach improves the quality of information retrieved for GenAI Chat, RAG, and Agentic RAG.

Reranking

Retrieved documents or document fragments can be processed by a dedicated reranker model.

The reranker evaluates the relevance of candidate results and reorders them before the information is provided to the LLM.

This improves the quality of the context used for AI-generated responses.

Image-Based File Indexing

Image-based files can participate in AI indexing and retrieval.

Depending on the configured processing pipeline, elDoc can:

  1. process the image using OCR or a vision-language model;

  2. extract textual or structured information;

  3. create search and vector indexes;

  4. make the extracted content available for semantic retrieval, GenAI Chat, RAG, and other AI capabilities.

This allows scanned archives and other image-based repositories to participate in the same AI knowledge environment as digitally generated documents.

AI and Document Security

AI functionality operates within the elDoc access-control model.

When information is accessed through:

  • GenAI Chat;

  • semantic search;

  • RAG;

  • Agentic RAG;

  • AI Agents;

  • AI Indexing;

document permissions remain enforced.

A user can retrieve or use document content through AI functionality only when that user has the required access rights to the underlying document.

For RAG and Agentic RAG, only content from documents accessible to the current user is included in the retrieval process.

Local and Cloud AI Models

elDoc supports both cloud-based and locally deployed AI models.

Depending on the organization's configuration, AI processing can use:

  • cloud AI providers;

  • AWS Bedrock;

  • OpenAI-compatible APIs;

  • locally hosted models;

  • private or isolated AI infrastructure.

This allows organizations to select the AI deployment model that best matches their security, privacy, performance, cost, and regulatory requirements.

Specialized Models

Different AI tasks can use different specialized models.

elDoc can use separate models for functions such as:

  • Chat;

  • reasoning;

  • AI Agents;

  • tool usage;

  • embeddings;

  • reranking;

  • AI Document Processing;

  • OCR and vision processing.

This allows each AI operation to use a model appropriate for the corresponding task.

AI Across the elDoc Platform

AI functionality is integrated across multiple elDoc modules rather than operating as a separate standalone component.

AI capabilities can be used in:

  • AI File Management — Chat, semantic retrieval, AI Indexing, AI Agents, and document analysis;

  • AI Document Processing — OCR, classification, extraction, and document understanding;

  • Document Workflow Automation — AI-assisted processing and automation;

  • Enterprise search and knowledge retrieval — semantic search, RAG, Agentic RAG, and reranking.

Last modified: August 26, 2026