Acquit.ai

The disciplines of engineering
applied to the art of law.

The failure-prevention engineering (FMEA, Lean Six Sigma, defense-in-depth) that runs hyperscale cloud and aerospace systems. Adapted for adversarial human systems.

The Stack

Three integrated layers. One coordinated system.

Layer 1 · Evidence

ExhibitCTL

MIT open source · runs on your hardware

The evidence pipeline. Multi-source collection, SHA-256 chain of custody at point of capture, AI-powered classification, and court-ready exhibit packaging. The foundation everything else rests on.

Layer 2 · Discovery

FrawdBot

Source-discovery / forensic-discovery agent

Surfaces new evidence sources and anomaly patterns from Google Workspace admin logs. Insider-threat signals, OAuth anomalies, spoliation patterns, and bulk-download spikes. All turned into evidentiary leads that feed Layer 1.

Layer 3 · Strategy

Wargame Engine

Adversarial coordination layer

Helps teams reason across actors, incentives, constraints, and likely moves in a multi-party dispute. The Acquit Score™, Sentient Analysis, ARPN risk scoring, and FMEA-grade method live here.

"Information dominance against fractured opponents."

"The system reads the full game state. Each opposing participant usually sees only a slice."

The Problem

Litigation intelligence has a distribution problem.

500K

solo and small law firms

2.5M+

active commercial litigation cases

86%

of civil winners receive < $250K

75%

of civil cases include a pro se party

$1.8M

median eDiscovery spend (RAND)

The tools exist. The access doesn't. Until now.

The Distinction

One word. Entirely different intelligence.

Sentiment Analysis

  • Reads document tone
  • Flags positive / negative / neutral
  • Backward-looking: what was said
  • Surface-level NLP keywords

Sentient Analysis

  • Models cognitive architecture
  • Maps 5-tier behavioral modifiers
  • Forward-looking: what they'll do
  • Agent-level behavioral prediction

The Gap

  • Tone ≠ action
  • Keywords ≠ prediction
  • Description ≠ intelligence
  • Reading transcripts ≠ understanding minds

The Evidence Pipeline

Six capabilities feed the sentient analysis engine.

FMEA is how engineers prevent failures in chipsets, data centers, and rockets. When NASA confirms a catastrophic event, the first order is: lock the doors, preserve everything. That's right of boom. We start left of boom. Your evidence base is collected, hashed, and court-ready before the lawsuit lands.

01

Collect

Evidence collection starts before litigation, not after. ExhibitCTL applies the same failure prevention methodology used in manufacturing, cloud infrastructure, and aerospace (FMEA) to digital evidence. Multi-source, defense-in-depth. If one method is challenged, others back it up.

02

Verify

SHA-256 chain of custody hashing at creation, never retroactive. Automated verification with cryptographic audit trails. Every exhibit is independently verifiable.

03

Classify

AI-powered evidence classification by evidentiary strength. Your strongest material surfaces first, so your attorney focuses where it matters most.

04

Brief

Attorney-ready briefings with severity-tiered findings. Scannable in 60 seconds, deep-readable when your lawyer has time. Evidence-ranked, action-oriented, court-grade.

05

Present

Court-ready PDF exhibits with embedded metadata. Attorney-approved packaging with exhibit guides. Delivered as dated, hash-verified bundles your counsel can file directly.

06

Prove

Cryptographic derivation proofs linking every trimmed exhibit to its authenticated source. Opposing counsel challenges your evidence? Mathematical proof it's unaltered.

The Engine

The Sentient Analysis Engine

Evidence in. Predictions out.

Evidence Pipeline

Collect

Verify

Classify

Brief

Present

Prove

Behavioral Modifier Stack

T1 Belief Architecture
T2 Motivation Stack
T3 Emotional State
T4 Cognitive Impairment
T5 Coordination Dynamics

91%

Acquit Score™ calibration

201

tracked predictions

5

behavioral modifier tiers

Trust Architecture

Your evidence never leaves your hardware.

Privacy is the #1 barrier to attorney AI adoption. We designed around it. ExhibitCTL runs locally: no cloud, no vendor access, no third-party exposure.

SHA-256 hashing

At point of collection, never retroactive

Local-first architecture

Analysis runs on your hardware

Open-source core

Inspect every line of code

Open Source

Built on ExhibitCTL. Free to collect. Expert services when you need them.

An open-source evidence collection pipeline. MIT licensed. Your evidence stays on your hardware. When you need deeper analysis, expert witnesses, or managed eDiscovery, we're here.

Why This Matters

The courthouse has a cover charge.

86% of civil trial winners receive awards under $250,000. Millions of cases per year fall in a range too small for institutional funders but too expensive to fight with traditional staffing. AI changes the math.

Read: The Litigation Dead Zone →

Who This Is For

Two problems. One pipeline.

For Law Firms

200GB of evidence. 72 hours to review.

ExhibitCTL collects in hours, not weeks. Sentient analysis prioritizes by evidentiary strength. Your strongest material surfaces first, so your team focuses where it matters most.

For Business Owners

Your own data. Finally making sense.

Turn raw data into court-ready exhibits. Emails, documents, and digital records, collected with chain of custody, classified by relevance, and packaged for your attorney.

Request an Invite

Acquit.ai is invite-only. We're starting with activist plaintiff-side firms, mission-aligned funders, and people doing the work that the dead zone leaves behind. The application is short. We read every one.

Or just stay close. Drop your email for occasional methodology updates while you decide whether to apply.