Documentation & Manual

Cleros AI Platform Guide

Master the complete litigation intelligence workflow: case setup, multi-document ingestion, 7-agent DAG execution, and simulated jury deliberations.

Standard Litigation Workflow

Step 01
Case Creation & Venue Demographic Selection
Initialize your matter by selecting jurisdiction (federal district or state and county venue). Cleros AI instantly queries local ground-truth demographic profiles, including educational levels, median household incomes, and political distributions from US Census databases.
Step 02
Document Ingestion & Automated OCR
Upload multi-party complaints, answers, contracts, expert disclosures, and deposition transcripts. Our pipeline performs high-speed OCR, Bates indexing, and semantic chunking while preserving line numbers and exhibit cross-references.
Step 03
Autonomous 7-Agent Legal DAG Reasoning
The system triggers the seven legal agents in sequence: Fact Timeline Extraction, Entity Graphing, Element Auditing, Evidence Triangulation, Contradiction Discovery, Liability Assessment, and Strategy Memo Generation.
Step 04
12-Person Deliberation Simulation
Instantiate a 12-person virtual jury reflecting your chosen venue. Jurors elect a foreperson, deliberate evidence, allocate comparative fault percentages, and vote across multiple deliberation rounds to predict verdict probabilities.
Step 05
Voir Dire & Strike Plan Generation
Review customized 4-tier questioning sequences and cause challenge lock-in templates. Identify prospective juror risk scores and export a printable courtroom strike sheet.
Step 06
Court-Ready PDF Intelligence Export
Generate an executive trial strategy dossier formatted for managing partners, clients, or mediation presentation binders with full citation indexes and damages exposure curves.

Supported Ingestion Formats & Confidentiality

Cleros AI natively processes scanned PDF filings, native Microsoft Word (.docx), ASCII/PTX deposition transcripts, plain text productions, and scanned exhibit images. All files undergo OCR indexing in isolated, single-tenant memory spaces protected by AES-256 encryption. Client files are never retained in persistent shared memory and are never utilized to fine-tune public models.

Deep Dive Into Features

Ready to Put the Platform to Work?

Start your first case analysis with 1 complimentary trial credit.