Connecting Validis to your AI client (Admin guide)

This guide is for IT or workspace administrators setting up the Validis MCP connector for users in Claude, ChatGPT, or Microsoft 365 Copilot. Once configured, users in your workspace can query Validis data directly from their AI client using their own Validis credentials.

Looking for the end-user guide instead? See Using Validis with AI tools.

Server details

Authentication

Regional endpoints

Connect directly to the Validis instance for your region:

A regional endpoint will only authenticate users whose Validis accounts live in that region.

Prerequisites

The following apply regardless of which AI client you use. Additional client-specific requirements are listed under each client below.

• Each end user must have valid Validis portal credentials for the Validis instance they'll use
• Outbound HTTPS access to *.validis.com from your network

Setup by AI client

Every setup follows the same pattern: a workspace admin creates the Validis connector once for the whole workspace, then each end user authenticates individually with their own Validis credentials. End-user authentication grants the AI client access only to the engagements that user already has permission to see in Validis.

Claude (Claude for Work or Claude Enterprise)

Additional prerequisites

• A Claude for Work or Claude Enterprise workspace
• Workspace admin (Owner) permissions in Claude
• Custom connectors must be enabled at the organisation level by the primary owner. If they aren't yet enabled for your organisation, contact your Claude administrator.

Step 1: Create the connector

• Sign in to Claude as a workspace admin (Owner).
• Open Settings, then go to the workspace administration area.
• Locate Connectors (or Apps and connectors) and select Add custom connector.
• Enter the following: 
  • Name: Validis  
  • Description: Access to Validis accounting and audit data
  • URL: the regional endpoint from the table above
  • Authentication: OAuth 2.0
• Save the connector.

Step 2: Enable the connector for users

• After saving, confirm the connector is enabled or published for the workspace. Depending on your Claude plan, this may happen automatically on save, or may require an explicit publish/enable step.
• If your organisation has multiple workspaces, confirm the connector is available in each workspace where it should be used.

For more on custom connectors in Claude, see Claude Support: Get started with custom connectors using remote MCP.

ChatGPT (Business or Enterprise)

Additional prerequisites

• A ChatGPT Business or ChatGPT Enterprise workspace
• Workspace owner or admin permissions in ChatGPT

Step 1: Enable developer mode and create the connector

• Sign in to ChatGPT as a workspace owner or admin.
• Open Workspace Settings, then Permissions & Roles.
• Locate Developer mode / Create custom MCP connectors and enable it. If your workspace uses role-based access, you can restrict this to selected administrators.
• Open ChatGPT and go to the Apps/Connectors area. Choose Create App or Add Custom Connector.
• Toggle Developer mode on and acknowledge the warning that custom connectors run third-party code.
• Select Create app (or Add custom connector).
• Enter the following: 
  • Name: Validis 
  • Description: Access to Validis accounting and audit data
  • Server URL: the regional endpoint from the table above
  • Authentication: OAuth 2.0
• Save the connector.
• Publish or share the connector with the workspace so it's available to users. If your workspace uses role-based access controls, assign it to the appropriate users or groups.

Step 2: End users authenticate

• In ChatGPT, open Settings, then Connectors.
• Locate Validis in the list and select Connect.
• Sign in to Validis when prompted. The OAuth flow returns you to ChatGPT once complete.
• In a new chat, select the + icon and enable Validis for the conversation.
• Reference the connector explicitly when prompting, for example: "Using Validis, list the engagements I have access to."

For more on MCP apps in ChatGPT, see OpenAI Help: Developer mode and MCP apps in ChatGPT.

Microsoft 365 Copilot and Teams

Validis is made available in Microsoft 365 Copilot and Teams as an agent that connects to the Validis MCP server. A workspace administrator builds and publishes the agent once for the organisation (Part 1). Each end user then adds the agent and signs in with their own Validis credentials, seeing only the engagements they already have access to (Part 2).

Important: Retrieving Validis data through this agent requires a Microsoft 365 Copilot add-on license (or Microsoft 365 E7). Users on Copilot Chat / Copilot "Basic" without the add-on cannot retrieve data from the connector, even after an administrator sets it up correctly. Confirm your intended users hold the required license before rollout.

Prerequisites

• Microsoft 365 Copilot add-on licensing (or E7) for the users who will use Validis
• Access to Microsoft Copilot Studio with permission to create, publish, and share agents
• Each end user must have valid Validis portal credentials with access to at least one engagement

Part 1: Build and publish the Validis agent

• Sign in to Microsoft Copilot Studio.
• Select Create and build a new agent. Give it a name (e.g. Validis) and a short description.
• Open the agent's Tools tab, select Add a tool → Add a new tool → Model Context Protocol.
• Enter the details below:

• Field	Value
  Server name	Validis
  Server description	Access to Validis accounting and audit data
  Server URL	The regional endpoint for your users (see table above)
  Authentication	OAuth 2.0
  Type	Dynamic discovery

• Continue, and create the connection when prompted, signing in to Validis to authorise it. (This authorisation is for setup only. End users authenticate separately with their own credentials.)
• In the tool's configuration, under Additional details, confirm "Credentials to use" is set to End user credentials. This ensures each user authenticates with their own Validis login and only sees their own engagements.
• Confirm the tool shows Enabled and save the agent.
• Select Publish.
• Open Share settings and share the agent with the appropriate users or the whole organisation.
• On the Channels tab, make the agent available on Microsoft 365 and Microsoft Teams (and SharePoint if required), and publish the channel availability.

Note: If the tool shows a connection or "Couldn't retrieve" error immediately after setup, wait a short time and retry. The connection can occasionally need a moment to become ready.

Part 2: End users access Validis in Copilot or Teams

Requires a Microsoft 365 Copilot add-on license. See the licensing note above.

• In Microsoft 365 Copilot or in Microsoft Teams, find the Validis agent in your agents list. In Teams, look under Apps, where it may appear under "Built for your org." Newly published agents can take some time to appear.
• Open the agent and ask a question, e.g. "Fetch me a list of my Validis engagements."
• The first time you use it, the agent prompts you to connect. Select Open connection manager, sign in with your own Validis credentials, and once connected, select Retry. The agent only accesses the engagements your Validis account already has permission to see.

For more on MCP servers with Copilot agents, see Microsoft Learn: Extend your agent with MCP.

Verifying access

A successful response listing your engagements confirms the connection is working. If no engagements are returned:

• Check the user's account holds a Microsoft 365 Copilot add-on license (Copilot/Teams only)
• Check the Validis account has access to at least one engagement in the relevant region

Privacy

Your Validis credentials are never shared with the AI tool. You sign in directly with Validis through a secure OAuth flow.

For full details, see our Privacy Policy.

Support

For setup or connection issues, contact your Validis representative or visit Validis Support.

See also: Using Validis with AI tools. Our end-user guide for connecting and querying Validis once your admin has completed the setup above.