AI Frontier · Run in Agent
Microsoft Copilot — the platform inside Microsoft 365
The platform ships as a declarative agent for Microsoft 365 Copilot. It installs inside the Copilot and Teams your organization already uses, so your team asks about members, campaigns and metrics — or runs an approved operation — without a new tool to learn. It connects through the platform's governed connection with a tenant key, under the same scopes, audit trail and budget as every other way you use the platform.
How it works
Four steps to run the platform inside Copilot
Install the declarative agent in Microsoft 365 / Teams
Add the platform's declarative agent to the Copilot and Teams your organization already runs. There's no separate app for your team to adopt — it shows up where they already work.
Connect with your tenant key
The agent links to your workspace through the platform's governed connection using a tenant key. One connection, owned by your organization, controls what the agent can see and do.
The governed tools appear in Copilot
Your approved platform tools become available inside Copilot. Each tool carries the same per-tool permissions as everywhere else, so Copilot can only reach what your team is already allowed to reach.
Ask a question or run an approved operation
Ask about members, campaigns and metrics in plain language, or trigger an approved operation — right in the Copilot window, without switching tools or learning a new interface.
What your team gets
The platform where your team already works
Answers in the Copilot they already use
Your team asks about members, campaigns and results inside Microsoft 365 Copilot — the assistant they open every day. Nothing new to install on their side, nothing new to learn.
Approved operations without leaving Microsoft 365
Beyond reading, the agent can run operations you've approved — so common tasks happen inside Copilot instead of a separate console. Write actions stay gated by the same permissions as the rest of the platform.
Grounded in the semantic layer
Every number comes from the certified semantic layer, so the figure Copilot returns matches your dashboard. The AI reports the metric it's given — it never invents a number.
The same governed tools are exposed through the MCP layer, so Copilot is one more way to reach the platform under a single set of rules.
Governance
One connection, the same guardrails
Running inside Copilot doesn't open a side door. Every call uses the same tenant key, scopes, audit trail and budget as the rest of the platform.
Tenant key
The agent connects through one governed connection tied to your workspace. Your organization owns the tenant key and can rotate or revoke it at any time.
Per-tool scopes
Each tool exposed to Copilot carries its own scopes. The agent can only reach the data and actions those scopes allow — the same permissions you use everywhere else.
Read vs. write
Questions and lookups are read-only. Any operation that changes data is a write action, gated separately so nothing runs unless it's been approved.
Audit trail
Every call the agent makes is recorded on the same audit trail as the rest of the platform, so you can always see who asked for what and what ran.
Budget
Usage draws on the same budget as every other way you connect, keeping cost and volume under one governed ceiling instead of a separate, unmanaged channel.
The platform inside the Copilot your team already uses.