All posts

Synced or Federated: Connecting Business Apps to Copilot

A useful Copilot answer appears in seconds. Afterwards, the user often still opens another application, finds the record and carries the work forward by hand. Connected apps such as Adobe Express, Figma, Dynamics 365, Box, monday.com and Miro now appear directly inside Microsoft 365 Copilot. Copilot connectors link more than 100 external sources, either synced into Microsoft Graph or federated in real time. In this article, I explain the difference between synced and federated connectors and which four questions leaders should answer before connecting more systems.

Answer quality covers only part of enterprise AI. The larger friction point often follows the answer: the user opens another application, finds the right record, recreates the context and carries the work forward by hand. Microsoft's move to bring business apps directly into Microsoft 365 Copilot reduces the distance between insight and execution.

Context switching as the bottleneck

A marketer receives a strong campaign concept in Copilot and then switches to a design platform. A seller reviews an opportunity summary and then updates the CRM. The AI answer is useful, but the workflow breaks. Microsoft names this gap directly: the insight lives in one place, while the action belongs somewhere else.

From assistant to work surface

Agents connected to Adobe Express, Figma, Optimizely, Dynamics 365, Box, monday.com, Miro and Wix show interactive experiences inside Copilot. Users preview content in context, review or update records at the moment of decision, create visual assets without leaving the flow and see project information where the conversation takes place. Every additional click breaks continuity and increases the chance that a task stalls.

Synced and federated connectors

According to Microsoft Learn, Copilot connectors extend Copilot and Microsoft Search to external data. Microsoft distinguishes two types:

  • Synced connectors index content into Microsoft Graph.
  • Federated connectors fetch data in real time without indexing it into Microsoft 365.

Some scenarios benefit from indexed content that is discoverable across search and Copilot. Others require live access to dynamic or sensitive data that should remain in the source system. Microsoft documents more than 100 connectors for Microsoft services, SaaS platforms, business systems and on-premises sources. Copilot thus becomes an access layer across enterprise knowledge and workflows.

A change in the operating model

With connected apps, employees use AI to move work through systems instead of only asking it for help. Marketing teams get from idea to asset faster. Sales teams spend less time on CRM administration. Operations teams act on recommendations in context. Project teams update work items without losing the thread. In each case, the main gain is continuity, which improves usability and therefore adoption.

Governance grows with connectivity

More connected apps and data sources require clear rules: which apps are approved, which data sources are connected, how permissions apply, what is indexed and what is fetched live, and how administrators monitor these experiences. Microsoft designs the connectors to be permission-aware, so users only see content they may access in the source system. A connected Copilot affects information architecture, app governance, identity, compliance and change management, not only the front end.

Four questions for leaders

Where does the workflow break today? Places where employees receive useful AI output but re-enter or reformat information in another system offer the greatest value.

Which systems matter most? A broad app catalog is not the goal. Priority goes to the systems at the center of execution, such as CRM, service management or project management.

What should be indexed, and what should stay live? Stable content suits synced connectors. Sensitive or fast-changing data needs federated access.

How do employees learn to work differently? If teams keep treating Copilot as a detached chatbot, the technology underperforms. The opportunity lies in redesigning work so that insight and action follow each other without interruption.

The winners in enterprise AI will be those who embed AI into real work across real systems. Connected apps move Copilot toward an operational layer for daily work.