Technical details

How the Microsoft Programs Assistant works

It works in Microsoft Teams. An AI model reads the question and calls one of three read-only tools on our MCP server, which reads Microsoft Partner Center. Every fact comes from tested code, so the same question gets the same facts every time.

Try Demo
Watch Demo
Test with your own MCP client
Tools
Architecture

// .vscode/mcp.json
{
  "servers": {
    "microsoft-programs-assistant-demo": {
      "type": "http",
      "url": "<DEMO-URL>",
      "headers": { "x-api-key": "<TEST-KEY>" }
    }
  }
}

curl <DEMO-URL> \
  -H "x-api-key: <TEST-KEY>" \
  -H "content-type: application/json" \
  -H "accept: application/json, text/event-stream" \
  -d '{"jsonrpc": "2.0", "id": 1, "method": "tools/call",
       "params": {"name": "portfolio_insights", "arguments": {"product": "Intune"}}}'

# pip install "mcp>=2.3"
import asyncio

import httpx2
from mcp.client import Client
from mcp.client.streamable_http import streamable_http_client

URL = "<DEMO-URL>"
KEY = "<TEST-KEY>"


async def main():
    http = httpx2.AsyncClient(headers={"x-api-key": KEY})
    async with Client(streamable_http_client(URL, http_client=http)) as client:
        result = await client.call_tool("portfolio_insights", {"product": "Intune"})
        print(result.content[0].text)


asyncio.run(main())

At a glance

Runs in: Microsoft Teams, as a Copilot Studio agent
Who can use it: MDS staff we invite, signed in with Microsoft Entra ID
AI model: Claude, provided by Microsoft in Copilot Studio. The live demo uses Claude on Microsoft Foundry
Tools: 3 tools on our MCP server, all read-only
Data: Microsoft Partner Center: the incentive catalog, our claims per client, and payment statements
Hosting: Azure App Service, run by MDS

How it answers a question

A person asks in Teams, signed in with Microsoft Entra ID. The agent’s AI model reads the question.
The model picks a tool by its name and description, and calls it on our MCP server. MCP (Model Context Protocol) is an open standard that lets an AI model call tools on a server.
The tool reads Microsoft Partner Center, applies our business rules, and returns facts and warnings. The model writes the answer from them and keeps every warning.

The AI model never decides a fact or a number. Those come from code that has unit tests. The model only picks the tool and words the answer.

Architecture

Figure 1. Who runs what. Steps 1–6 follow one question. Microsoft runs Copilot Studio and its own services. MDS runs the MCP server. Dashed: what only people see, in the Partner Center portal.

Figure 2. The same question, end to end. The agent can call tools more than once, in a loop, before it answers.

Figure 3. The live demo. Steps 1–4 follow one question. The demo server runs the real agent’s code on made-up test data, and Claude on Microsoft Foundry picks the tool and writes the answer. The demo has no connection to Partner Center. The icons in all three figures are Microsoft’s official architecture icons.

Tools

Three tools, each with one job. Each one is marked read-only, and the server has no tool that sends, claims or changes anything.

Tool	Ask it	Inputs	What it returns	Limits
`list_programs`	“Which programs pay on Copilot?”	None	Every program in Microsoft’s incentive catalog, with the products it pays on	The catalog doesn’t show which programs we’re enrolled in
`customer_incentives`	“What can we still claim at Northwind Traders?”	`customer`: a name or tenant ID. Or `products`: a prospect’s products, separated by commas	The products we haven’t claimed yet at that client, and the one to offer next	When several clients match, it lists them, and the agent asks which one
`portfolio_insights`	“Who should we take Intune to?”	`product`, optional	Without a product: the biggest gaps across all clients. With one: the clients without a claim for it, most other claims first	Eligibility isn’t in the data, so every answer says to check it in Partner Center

Test it yourself

Both ways use our demo server. It runs the same code and tools on made-up data for Trey Research, a made-up partner. No real client data is on it.

In your browser

Ask the live demo your own questions. A real AI agent picks the tool and writes the answer.

Try Demo

With your own MCP client

We send you the demo server’s address and a test key. Put them in place of <DEMO-URL> and <TEST-KEY> in the samples at the top of this page. Any MCP client that can send a header works, for example VS Code.

Ask for a test key

Ask the questions it was made for

Which incentive are we missing out on the most?
Who should we take Intune to?
What can we still claim at Northwind Traders?
A prospect uses Exchange Online, Intune and Power BI. What could we claim?
Which programs pay on Copilot?

You askWho should we take Intune to?

The model callsportfolio_insights {"product": "Intune"}

The tool returnsIntune: Trey Research has claimed it at 21 of 32 customers.
Customers without a Trey Research claim for it, most other Trey Research claims first (11 in all): Tailwind Traders (11), Adventure Works Cycles (9), Margie's Travel (7), Fabrikam, Inc. (5), Lamna Healthcare Company (5)
Eligibility isn't in the data. Check it in Partner Center before you contact the customer.

The model then writes a short answer from this text, and keeps the warning.

	In Teams	Demo server
Code, tools and rules	Ours	The same
Data	Microsoft Partner Center	Made up, for Trey Research
Who can use it	MDS staff we invite, signed in	Anyone, in the live demo. With your own client: anyone with a test key
AI model	Claude, in Copilot Studio	Claude on Microsoft Foundry in the live demo, or the model in your MCP client
Key	Private	A test key from us

Limits

It only reads. No tool can enroll, claim or change anything, so a person makes every decision.
Eligibility isn’t in the data. Our consultants check it, and that check will be added to the assistant later.
The demo answers a limited number of questions each day, and we can change test keys at any time.

Want one for your team?

Tell us how your team works today, and we’ll show you what this assistant could do.

Try Demo Talk to us