> ## Documentation Index
> Fetch the complete documentation index at: https://labs.laer.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# System card

**Epiq AI** is a GenAI platform designed to support secure, scalable, and defensible workflows for litigation tasks such as document review, fact discovery, and knowledge retrieval.

***

## Core Capabilities

**1. AI-Driven Document Review**

* Identifies responsiveness, privilege, and issue relevance
* Supports multiple issues simultaneously
* Tailored to client protocols with minimal prompt engineering

**2. Integrated Chat Q\&A**

* Allows case teams to ask natural language questions across document sets
* Answers are grounded in retrieved evidence and tailored to matter-specific knowledge
* Ideal for early case assessment, investigations, and fact discovery

**3. Custom Knowledge Layer**

* Structured knowledge entities from diverse sources (documents, facts, prior knowledge)
* Instructs the model through transparent, editable prompts
* Promotes consistency and auditability in outcomes

**4. Integrated AI Orchestration**

* Modular architecture for chat, search, and classification
* Flexible tuning for task-specific prompts
* Human-in-the-loop mechanisms embedded across flows

***

## Key Components

| Component              | Functionality                                                                                                                                                     |
| ---------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Case Knowledge**     | Entity-centric knowledge (e.g., persons, organizations, relations) extracted from case documents                                                                  |
| **Protocol Knowledge** | Transform user-defined protocols into background knowledge, issue scopes, and instructions.                                                                       |
| **Chat Q\&A Engine**   | Natural language interface for retrieving and synthesizing answers from case documents                                                                            |
| **Search & Retrieval** | Multi-index search using metadata, extracted text, embeddings, and structured knowledge                                                                           |
| **Classification**     | Predict document-level tags (e.g., responsiveness, privilege, issues) with relevancy scores                                                                       |
| **Extraction**         | Extract and summarize information from documents                                                                                                                  |
| **Workflow**           | Purpose-built workflows by combining classification, extraction, reasoning, and validation into end-to-end AI workflows (e.g., privilege review, PII/PHI review). |

***

## Models

<table><thead><tr><th>Name</th><th width="250">Capabilities</th><th>Version</th><th data-hidden>Context (GPT tokens)</th></tr></thead><tbody><tr><td><strong>GPT-5.4</strong></td><td>Chat</td><td>2026-03-05</td><td>272K tokens</td></tr><tr><td><strong>Llama 3.1 70B</strong></td><td>Chat</td><td>llama-3.1-70b-turbo</td><td>32K tokens</td></tr><tr><td><strong>AIDA Reasoning Model</strong></td><td>Workflow, Search, Classification, Extraction</td><td>2.0</td><td>128K tokens</td></tr><tr><td><strong>AIDA Extraction Model</strong></td><td>Search, Extraction</td><td>2.0</td><td>128K tokens</td></tr><tr><td><strong>AIDA Embedding Model</strong></td><td>Search, Classification</td><td>1.0</td><td>-</td></tr><tr><td><strong>Whisper Large V3</strong></td><td>Transcription</td><td>large-v3-turbo</td><td>-</td></tr></tbody></table>

## Security & Governance

* Secure cloud deployment options
* Full audit trail and role-based access control
* Model behavior can be reviewed and adapted by case teams

***

## Extensibility

* Supports API integration with Relativity and Epiq Discovery
* Protocols and prompts are user-editable and version-controlled
* Custom classification model and knowledge management per matter or client

***

## Use Cases

* Responsiveness/Issue Classification
* Privilege Classification
* Fact Discovery & Investigations
* Early Case Assessment
* Deposition Preparation