Everlaw

Everlaw provides a secure, standardized Model Context Protocol (MCP) server that allows OpenAI to connect to Everlaw and work with live matter data on behalf of an authenticated user.

Through the connector, a connected agent can understand a user’s Everlaw project structure, translate natural-language requests into Everlaw search logic, retrieve responsive documents and document text, and analyze metadata and Bates driven result sets. The connected agent can also support natural-language reporting and analysis workflows, while Everlaw remains the governed evidence layer and system of record for case materials, metadata, and work product.

Access to Deep Dive via this connector complements search and retrieval tools, as well as Everlaw’s native Deep Dive experience. It is intended to support broader, semantic, evidence-backed questions while preserving Everlaw’s permissions and the integrity of the underlying matter record.

The connector is designed to operate within the user’s existing Everlaw permission level. Every request is authenticated and authorized against the user’s Everlaw permissions, so the agent can access only the projects, documents, and data that user is already permitted to view. Deep Dive will also be available through the connector, allowing a user to query Deep Dive over eligible Everlaw project data and retrieve evidence-backed results without changing the underlying matter record.

Using the connector, an agent can:

Use Deep Dive to ask broader investigative questions and return evidence-backed results, once the Deep Dive tools are available.

Convert a natural-language request into a structured Everlaw search.

Generate structured reports about project contents, document volumes, custodians, dates, Bates ranges, and other returned data.

Search by document text, metadata, Bates or Control number, binder, coding, redactions, productions, processing criteria, and other supported terms.

Identify the metadata fields, datasets, and processed uploads available in a project.

Retrieve matching documents with Bates numbers, metadata, review links, document-text links, and extracted values.

Summarize and analyze retrieved documents or result sets in natural language.