---
title: "Copilot agent in the Microsoft Copilot Store | Digital Cuisine"
description: "Copilot agent with Microsoft Graph Search Connectors, large-scale Python data preparation and a custom SharePoint search for an international provider of market and company data."
url: "https://cuisine.digital/en/copilot-store-agent"
lang: en
generated: cuisines-markdown-v1
---

AI reference / Copilot agent in the Microsoft Copilot Store

# A Copilot agent you could buy

For an international provider of market and company data, we developed and integrated our first Copilot agent that could be purchased commercially in the Microsoft Copilot Store: with Microsoft Graph Search Connectors, Python data preparation and a custom SharePoint search.

[Request an initial AI conversation](https://cuisine.digital/en/contact?thema=KI-Beratung)

Technology in the project

-   Copilot Studio
-   Microsoft Graph
-   Power Platform
-   SharePoint
-   Python

**Customer**

International data provider

**Services**

AI

**Period**

April to June 2025

**Our role**

Development and integration of a Copilot agent

**Technologies**

Microsoft Copilot Studio · Microsoft Graph Search Connectors · Python · Power Platform · Power Automate · SharePoint

## An agent you could buy in the Microsoft Copilot Store

When the customer, an international provider of market and company data, wanted to make its data available in Microsoft 365, there was no precedent: until then, Copilot agents had been internal tools. We turned this into a product that other organisations could purchase in the Microsoft Copilot Store and use in their own tenant.

That changes the requirements. An agent that is sold must set up reliably in other organisations' environments, keep its data sources clearly separated and remain understandable to administrators. We designed the architecture around exactly these requirements.

## Making data accessible to Copilot and search

The integration through Microsoft Graph Search Connectors is at the heart of the solution. It connects two ways of accessing the same data: answers in the agent and finding content across the Microsoft search environment.

Python processes retrieve the source data from the customer's REST APIs and prepare it for the Graph connector. Data ingestion maps each record, with its title, description and source link, to the search fields and is designed for large volumes of data: batch processing, retries for temporary errors and a repeatable import that also supports ongoing updates.

Custom search verticals allow targeted access to the connected sources. In SharePoint, we extended the Microsoft Search Experience with a custom search area that makes the same data accessible outside the chat.

## Deployment and moving between environments

We automated deployment so that the agent could be set up not just once but repeatedly. The connector, search verticals and schemas are created by scripts rather than through a series of clicks.

On this basis, the agent can be migrated between development, test and production environments without recreating the configuration manually. We also produced deployment guides for those responsible on the customer's side, documenting each step from tenant to approval.

## Integrating the agent into Microsoft 365

We coordinated the agent's integration into Microsoft 365 applications such as Word, PowerPoint and Teams so that the data is available where people actually work.

## Sharing knowledge with administrators and users

The result included enablement material for both sides: training documents for administrators and practical guidance on writing prompts for users. This allows the agent to be operated and developed further without us.

## Technical implementation

The agent was built in Microsoft Copilot Studio. A Graph connector imported the prepared data into the Microsoft search environment; Python processed the data from the REST APIs and mapped it to the search fields.

Power Platform and Power Automate complemented the integration. Custom search verticals and the additional search area in SharePoint made the connected sources accessible outside the chat too.

Period: April to June 2025.

## Relevant next steps

[AI consulting](https://cuisine.digital/en/services/ai-consulting.md)

Clarify your task and a suitable starting point together.

[All references](https://cuisine.digital/en/references.md)

Our work in AI, enterprise software, web and apps.

## Related projects

![Linde: Copilot governance & platform architecture](https://cuisine.digital/images/projects/linde-global-website-relaunch/logo.webp)

Linde

### Copilot governance & platform architecture

[View project](https://cuisine.digital/en/linde-copilot-governance.md)

![Energiekontor: AI search for internal SharePoint knowledge](https://cuisine.digital/images/logos/energiekontor.svg)

Energiekontor

### AI search for internal SharePoint knowledge

[View project](https://cuisine.digital/en/energiekontor-sharepoint-ai.md)

Financial services provider

Financial services provider

### Data protection for Copilot use

[View project](https://cuisine.digital/copilot-datenschutzpruefung-finanzdienstleister)

[Planning something similar? Discuss a project](https://cuisine.digital/en/contact?thema=KI-Beratung)

We will get back to you within one working day.

---

[Original page](https://cuisine.digital/en/copilot-store-agent) · [Contact and project enquiry](https://cuisine.digital/en/contact)