Setup Steps to Search Your Everlaw Projects Using Copilot

You can now connect the Everlaw MCP Server to Copilot, Microsoft's AI assistant, to securely search your Everlaw projects. This means that you can securely bring in your Everlaw evidence layer when you use Copilot for drafting or analysis.

In this article we'll cover:

• Guidance for Admins to deploy the Everlaw connector within Microsoft
• How to connect to Everlaw from within Copilot 
• Background details on how Everlaw and Copilot connect

Requirements

Pre-requisite: Before a Microsoft admin can deploy the Everlaw connector, an Everlaw Organization Admin must switch on the OAuth2 setting within Everlaw. Once it's switched on, the connection to Copilot can be enabled from within the Microsoft admin center.

To search Everlaw from within Copilot, you must have the following:

• A Copilot account within an enterprise-level Microsoft tenant that has deployed the Everlaw connector 
• An Everlaw account with access to at least one project: The searches and evidence you have access to within Copilot are limited to the project(s) and documents you have access to.

Data privacy and security

Your data privacy and security is our priority. Your use of Everlaw and our data privacy and security commitments to you are governed by our Terms of Service and Data Processing Addendum.

Your interactions within your Copilot environment are governed by your specific customer agreement with Microsoft, which also outlines Microsoft’s security and data privacy practices.

Connect Copilot to Everlaw

Connecting Everlaw to Copilot happens in three steps.

Step 1: Everlaw Organization Admin Enables OAuth2

This step must be done by an Everlaw Organization Admin. Instructions for how to do so are in the Organization Admin: Everlaw API article.

Step 2: Microsoft Admin deploys the Everlaw connector

Once OAuth2 is enabled in Everlaw, a Microsoft Admin can deploy the Everlaw connector. Follow the steps in Microsoft's Deploy connectors in the Microsoft 365 admin center article for instructions on how to do so.

Step 3: A user authenticates their Everlaw credentials

After a Microsoft admin has set up the Everlaw connector, any Copilot user within that Microsoft tenant with access to at least one Everlaw project can use it. Currently, MCP connectors (such as the Everlaw one) connect in the following Copilot experiences:

• Microsoft 365 Copilot Chat
• Copilot in Excel
• Researcher agent

To connect from within Copilot Chat:

1. In your Copilot Chat account, go to Settings > Sources.
2. In the Browse Sources menu, search for Everlaw.
3. Select the Everlaw app and then Continue to Everlaw. This will take you to your Everlaw login.
4. Log into your Everlaw account to autheticate the connection. Now you can search your Everlaw projects using Copilot.

See Microsoft's Federated Connectors Overview article for more details on authenticating your credentials and using enabled connectors. Once you connect, you can start using Copilot to search and analyze your data.

For users on projects in multiple organizations

Copilot's ability to search Everlaw projects is based on the owning organization's enablement of OAuth2 in Everlaw. If you have access to projects owned by multiple organizations, you will only be able to use Copilot to search the projects for which OAuth2 has been enabled at the Organization level. 

You only need to follow the steps to form the connection within Copilot once. When an additional organization enables OAuth2, the projects within that organization that you have access to will immediately become searchable within Copilot.

Enabled search terms

You can use Copilot to search in a project based on:

• Document contents (text)
• Binders 
• Bates/Control # 
• Terms in the Metadata category of the Everlaw search page, such as email participants, custodians, and dates
• Codes and freeform codes
• Productions
• Search Term Reports
• Extraction fields (though not the extracted values from the fields)

Within Copilot, you can write queries in natural language that reference these terms. Copilot will translate them into Everlaw syntax and return the results.

For more technical details on the Everlaw MCP server, see our Everlaw MCP Server article.

Best practice: Use tools to structure your searches

The following tools are enabled, and accessible from the Everlaw connector with Copilot:

• GetProjects: Lists projects the signed-in user can access and is typically the first step because other project-scoped tools require a numeric project ID.
• GetProjectMetadataFields: Lists searchable project-level metadata fields, including field name, internal ID, and format. 
• GetProjectProcessedUploads: Lists processed uploads for a project, useful for auditing ingestion history and correlating documents to upload batches. 
• GetProjectDatasets: Lists native datasets in a project so searches can be scoped to particular data collections. 
• GetProjectSearchTermReports: Lists Search Term Reports for a project
• GetProjectCodes: Lists codes in a project
• GetProjectFreeformCodes: Lists freeform code categories in a project
• GetProjectProductions: Lists the productions in a project
• GetProjectExtractionFields: Lists the extraction fields in a project
• GetProjectBatesPrefixes: Lists the Bates prefixes in a project
• GetProjectUsers: Lists the users in a project
  This is only available for Organization Admins and Project Admins
• AskDeepDiveQuestion and GetDeepDiveResponse: Combined, these let you ask Deep Dive questions and receive Deep Dive responses (for enrolled projects only)

To make use of the tools, we recommend the following steps when you start doing searches in a project:

1. Ask Copilot for a list of the Everlaw projects you have access to: What projects in Everlaw do I have access to?
   Copilot will list the projects you have access to.
2. Select a project and ask which binders are in the project: What binders are in Example Project- Number one?
   Copilot will list all the binders you have access to.
3. Ask it what processed uploads are in the project: What are the processed uploads in the project?  
   Copilot will list all the processed uploads.
4. Ask which datasets (native uploads) are in the project: What datasets are in this project?
   Copilot will list all the datasets.
5. Ask for the metadata fields in the project: What metadata fields are in the project?
   Copilot will list all the metadata fields in the project.
6. Ask for the Search term reports in the project: What Search Term Reports are in the project?
   Copilot will list all the Search Term Reports in the project
7. Ask for the codes in the project:What codes are in the project?
   Copilot will list all the codes in the project.
8. Ask for the freeform codes in the project: What freeform codes are in the project?
   Copilot will list all the freeform code categories and the field type (e.g. text, datetime, etc.) in the project.
9. Ask for the productions in the project: What productions are in the project?
   Copilot will list all the productions in the project.
10. Ask for the extraction fields in the project: What extraction fields are in the project?
    Copilot will list all the extraction fields in the project.
11. Ask for the Bates prefixes in the project: What Bates prefixes are in the project?
    Copilot will list all the Bates prefixes in the project.
12. [For Organization Admins and Project Admins] Ask for the users in the project: What users are in the project?
    Copilot will list all the users in the project.

In addition to being a useful reference for you to gather information about the project, which can inform your lines of questioning, asking these questions gives Copilot information that it uses to structure the queries it makes to Everlaw and its responses to subsequent questions. You can think of asking these questions as a form of prompting that gives Copilot an understanding of your project's structure, which will help it make more precise queries.

We recommend these questions to help get oriented with a project.

Supplemental information: How Everlaw and Copilot connect

This section provides a bit of technical background on the connection between Everlaw and Copilot. It's not necessary to understand this section before using Copilot to search your Everlaw documents, but if you're interested in some background information on how Copilot and Everlaw connect, read on.

The connection between Everlaw and Copilot is formed through a Model Context Protocol (MCP). MCPs have emerged as the standard tool that software companies use to securely connect AI tools with service providers like Everlaw. The Everlaw MCP server provides a secure, standardized, and reliable connection, with defined tools that allow the Copilot AI agent to search your Everlaw documents. At this time, the MCP tools that Everlaw has developed allow Copilot a read-only connection to Everlaw, meaning that you cannot take actions such as modifying, exporting, or deleting documents from within Copilot.

Because the MCP server is authenticated based on each individual user's Everlaw permissions, Copilot's responses never share information beyond what a user has access to within Everlaw.

Once a connection to the MCP server is established, you can ask Copilot your natural language questions, and Copilot will then:

1. Interpret the question to identify the information it needs from Everlaw.
2. Translate the question into Everlaw Search and send the query to Everlaw through the MCP.
3. Use the information received from Everlaw to answer your question. 

Because Copilot understands natural language and can synthesize across documents, your searches are not limited to pulling together a specific set of documents based on their characteristics. Instead, you can ask targeted, analytical questions to understand and draw insights from your data.

While Copilot does not use the API directly, the tools it has access to mirror those of the Everlaw Search API endpoints for retrieving binders, Bates information, metadata, and text (contents) searches.  

Developers can check out our Everlaw MCP Server article, which has technical details about our MCP implementation.