From court records to source-linked legal workOpen source on GitHub
Global litigation infrastructure

From courtrecords tolegal work.

CourtXAI starts with the signals litigation teams already trust: court records, filings, exhibits, hearings, and deadlines. Then it turns them into a source-linked case system that agents and lawyers can operate together.

Start a workspacePrivate beta · by invitationFollow the story
matter://court-record/intake
Court sync running

Litigation workspace

court record01
hearings14
next date11d
Court dataEvidenceTimelineDraft
  • Imported parties, counsel, filings, and hearing history
  • Linked 87 facts to exhibits and source pages
  • Routed review and drafting tasks through scoped legal runtimes
Court record importCase metadataParties and counselEvidence facetsSource pagesVerified timelinesDraft citationsDeadline controlBrowser agentsAudit exportCourt record importCase metadataParties and counselEvidence facetsSource pagesVerified timelinesDraft citationsDeadline controlBrowser agentsAudit export

The platform story

The promise is not that AI writes a paragraph. The promise is that CourtXAI makes the case itself computable: court data, evidence, timelines, deadlines, agents, and audit become one reviewable legal operating layer.

1court data in2evidence structured3legal work out

One matter, end to end

A legal story the product can prove

01

COURT SIGNAL

A court record becomes the starting point.

CourtXAI starts with the signals litigation teams already trust: court records, dockets, filings, hearings, and deadlines. Case data becomes structured matter context instead of another tab to check.

  • Court record sync
  • Parties mapped
  • Next hearing deadline
02

DOCUMENT INTELLIGENCE

The matter record becomes evidence.

Orders, exhibits, pleadings, contracts, notes, and messages move through OCR, classification, entity extraction, and source-aware analysis.

  • 312 files indexed
  • 87 facts linked
  • 26 exhibits processed
03

CASE THEORY

Facts become a verified timeline.

Dated events, contradictions, gaps, confidence, source documents, and key-event flags become a reviewable chronology instead of a hand-built spreadsheet.

  • Source pages attached
  • Contradictions flagged
  • Key events verified
04

ACCOUNTABLE OUTPUT

Drafts stay tied to the record.

Pleadings, counters, affidavits, memos, notices, and hearing prep are generated as reviewable work product with citations and human approval points.

  • Citations attached
  • Reviewer gates
  • Audit trail exported

Platform moats

The system underneath the story

Court ingestion

Court record lookup, import, sync, hearing history, parties, counsel, filings, and next-date deadlines.

Evidence graph

CASE/UCO observables, entities, relationships, markings, annotations, hashes, and confidence.

Workflow engine

Temporal-backed workflow definitions, registry actions, expressions, retries, and scatter-gather steps.

Accountable AI

Source-linked output, human review gates, workspace scoping, audit logs, and exportable records.

Capability-based legal runtimes

Every legal task runs in the right environment

Direct

Trusted actions

Fast in-process registry actions for deadlines, documents, transforms, notifications, and legal workflow steps.

registryfor_eachretry_until
Sandbox

Isolated code

Firecracker microVM execution for custom analysis code, dependency installs, file inputs, and artifacts.

E2Bpackagesartifacts
AI agent

Scoped reasoning

Case-aware agent sessions with allowed tools, allowed directories, and capability-style filesystem access.

case_idtoolssessions
Browser agent

Court portal work

Programmatic browser actions and AI-driven observe-act loops for court portals and other legal systems.

BrowserRpcKEDAportals

Start with one matter. Prove the operating layer.

Import a case, connect the record, build the timeline, route the work through the right runtime, and keep every draft accountable to source material.

Get started