Copilot Connectors: Sources, Access, and Work Context

Microsoft Copilot connectors, formerly called Microsoft Graph connectors, are the mechanism that brings data from outside Microsoft 365 into Copilot's search and response context. When a connector is configured, Copilot can draw on that external source when answering questions in Teams, Outlook, and other Microsoft 365 apps. Tools like fenbs keep shared project rules and follow-up tasks for any AI that supports its connection, useful when you work across multiple Copilot-enabled environments and other tools.

How Copilot connectors work

A connector indexes content from an external system, such as a Confluence wiki, a Salesforce database, or a Jira project, into the Microsoft Graph. Once indexed, that content becomes available to Microsoft Search and to Copilot when the user asks a question that the connector's data can answer. The connector does not give Copilot live access to the source system; it indexes a snapshot and updates it on a schedule.

Some connectors are built and maintained by Microsoft; others come from third-party vendors or are built by your organization using the Graph connector API. Check the Microsoft documentation for the current list of supported connectors for the systems you use.

Which data sources connectors can reach

Microsoft publishes connectors for a range of enterprise tools. Common examples include Confluence, ServiceNow, Salesforce, Jira, and file repositories such as Box and Dropbox. Your organization's IT team configures connectors at the tenant level, not at the individual user level. You cannot add a connector yourself unless you have the required administrator permissions.

• Wikis and documentation: Confluence, MediaWiki, and similar tools.
• Project and issue tracking: Jira, ServiceNow.
• CRM and business data: Salesforce.
• File storage: Box, Dropbox, Google Drive (via the connector for Google Drive).
• Custom sources: any system your organization builds a connector for using the Microsoft Graph connector API.

The connector indexes the text and metadata of items in the source, but the depth of indexing, what fields are captured, and how often the index refreshes vary by connector. Check the documentation for each connector to understand its scope.

How access is controlled

Copilot connectors inherit the user's permissions in the source system. If you cannot see a page in Confluence because you are not a member of that space, Copilot will not surface that page in its responses, even if the connector has indexed it. This permission trimming happens at query time, not at index time. The index may contain the content, but the response only includes what the current user is allowed to see.

This is a meaningful security property: a connector does not give every user in your tenant access to everything in the connected system. Permissions from the source are respected. However, it also means that a user who has never been given access to a Confluence space will not benefit from a Confluence connector for that space. Access must be granted in the source system first.

What connectors do not cover

Connectors make external content searchable and available for Copilot to reference in responses. They do not give Copilot the ability to take actions in the connected system. Copilot cannot create a Jira ticket, update a Salesforce record, or write to Confluence through a connector. Those actions require a separate integration, such as a Copilot extension or a Microsoft Power Platform flow.

Connectors also do not carry the team's ongoing project state: who is working on what, what was decided in the last meeting, or what rules apply to a piece of work. That kind of working context is separate from the content the connector indexes. A project wiki can describe a process, but it typically does not say where the current task stands or what the team decided this week.

See Microsoft Copilot agents for how Copilot can take actions beyond searching indexed content, and Copilot Studio for building custom agents with connectors.

Keep shared project rules available to every AI you use

If you use Copilot for some tasks and other AI tools for others, each assistant starts without knowing the rules and decisions the team agreed on unless you tell it. A connector can surface a wiki page describing the rules, but a wiki page is not the same as a structured record of the decisions your team made and when. The distinction matters when an assistant needs to know not just what the rule says but whether it is current and who decided it.

How to do this with fenbs

1. Create a free account and add a project for the work that spans your Copilot-enabled environment and other tools.
2. Write the team's standing rules as Decisions and rules. Include the person who decided each rule and the reason behind it.
3. Add a Links and listings entry for each connected system your team uses, such as the Confluence space or Jira project that Copilot connectors surface. This gives the next AI session a path to the source.
4. Connect a supported AI assistant from Integrations and ask it to read the project context, including rules and Where we left off, before it starts work.
5. When the team makes a decision in a meeting or a Teams thread, record it in Decisions and rules rather than leaving it in chat history where a new assistant cannot find it.

fenbs does not replace Copilot connectors. Its job is to keep the team's project decisions and task status in a form that any connected AI can read, independently of which Microsoft 365 tools are in use. For more on connecting GitHub Copilot to your projects, see GitHub Copilot app setup.