Specialized review agents, governed legal-data connections, litigation evidence, investigative review, and European legal AI investment

Five current developments in review agents, legal-data connectivity, AI litigation, investigative evidence, and legal AI infrastructure.

Edition
eDiscovery Decoder Newsletter

Sent September 24, 2026

Lead story

Casepoint Deploys Purpose-Built Agents for Relevance and Issue Coding

Casepoint released Relevance Determination and Issue Coding agents within its eDiscovery application, with sampling, accuracy metrics, audit trails, and human validation built into the workflow.

What happened
Casepoint released Relevance Determination and Issue Coding agents within its eDiscovery application, with sampling, accuracy metrics, audit trails, and human validation built into the workflow.
Why it matters
Applying specialized agents to relevance and issue coding moves agentic AI into core document-review decisions. Sampling, recall and precision measurement, recorded reasoning, and human checkpoints give review teams concrete controls to examine when evaluating defensibility and workflow impact.
Watch
Capabilities and performance controls are described in Casepoint's September 23 announcement and an eDiscovery Today report that reproduces the release. No independent accuracy results or matter-level benchmarks were published.

SourcesCasepointeDiscovery Today

Briefing item

MyCase Adds a Claude MCP Connector and Firm-Wide AI Tools

8am announced a Claude Model Context Protocol connector for MyCase, along with a firm-wide AI agent and an AI document-drafting feature planned for broader release in December.

What happened
8am announced a Claude Model Context Protocol connector for MyCase, along with a firm-wide AI agent and an AI document-drafting feature planned for broader release in December.
Why it matters
Connecting a practice-management system to an external AI interface can reduce navigation and reporting work, but it also makes permission enforcement, data boundaries, logging, and output provenance central implementation questions. The distinction between currently available connector functions and future native features is important for procurement planning.
Watch
The connector's action count, adoption, and planned release dates are company statements reported by LawNext on September 23. The Firm Agent and IQ Draft were prototypes at publication, and results can differ because the MCP route uses Claude while MyCase's native AI features use OpenAI models.

Sourceslawnext.com

Briefing item

Unsealed Filing Details Evidence in Publishers' AI Copyright Case

A newly unsealed summary-judgment filing in the news publishers' copyright litigation against OpenAI and Microsoft cites internal communications and advances claims involving scraping, paywalls, and removal of copyright-management information.

What happened
A newly unsealed summary-judgment filing in the news publishers' copyright litigation against OpenAI and Microsoft cites internal communications and advances claims involving scraping, paywalls, and removal of copyright-management information.
Why it matters
The filing shows how internal communications, technical collection practices, source metadata, and system-development records can become central evidence in AI copyright litigation. Discovery teams handling model-training disputes may need to preserve and connect communications with crawler configurations, data lineage, licensing records, and content-processing logs.
Watch
The cited assertions come from the plaintiffs' summary-judgment filing and have not been adjudicated. The eDiscovery Today report states that the defendants' summary-judgment request had been filed but was not yet public at the time of publication.

SourceseDiscovery Todaystorage.courtlistener.com

Briefing item

Everlaw Reports Newsroom Investigations Tripled Year Over Year

Everlaw said newsroom investigations supported through Everlaw for Good more than tripled year over year as journalists used litigation-style review technology for large document collections.

What happened
Everlaw said newsroom investigations supported through Everlaw for Good more than tripled year over year as journalists used litigation-style review technology for large document collections.
Why it matters
The reported growth illustrates that document-review workflows extend beyond formal litigation into investigations conducted under tight deadlines and constrained budgets. Newsroom use places the same emphasis on secure collaboration, source citation, access controls, and rapid fact validation that legal review teams face.
Watch
Participation, growth, matter, donation, and document-volume figures are Everlaw claims published September 23. eDiscovery Today discloses Everlaw as an educational partner and sponsor; the figures were not independently audited in the cited sources.

Sourceseverlaw.comeDiscovery Today

Briefing item

Noxtua Raises €100 Million as Legal Publishers Take Ownership

European legal AI company Noxtua raised about €100 million in a Series C round, with German legal publisher C.H. Beck becoming majority shareholder and Austrian publisher MANZ also investing.

What happened
European legal AI company Noxtua raised about €100 million in a Series C round, with German legal publisher C.H. Beck becoming majority shareholder and Austrian publisher MANZ also investing.
Why it matters
Publisher ownership can give a legal AI provider durable access to authoritative content as legal-data quality and licensing become key product differentiators. The transaction also demonstrates the scale of capital being committed to regionally governed legal AI alternatives in Europe.
Watch
Funding size, ownership changes, content strategy, and expansion plans are company statements reported by Artificial Lawyer and Legal IT Insider on September 23. The cited coverage does not disclose valuation, detailed commercial terms, or independent measures of product adoption.

Sourcesartificiallawyer.comLegal IT Insider

Educational only. Not legal advice. Review sources and matter-specific obligations before relying on any workflow decision.