Glossary
Microsoft Copilot refers to a series of AI-powered assistants from Microsoft. They help people with writing, analyzing, researching, presenting, and programming. The topic thus encompasses various AI technologies and areas of application. However, the data and features available depend on the specific Copilot product, the account used, and the license.
Contents
Copilot is an AI assistant that understands natural language and uses it to generate answers, text, images, analyses, or specific steps. Microsoft 365 Copilot is integrated into applications such as Word, Excel, PowerPoint, Outlook, and Teams. Within these applications, the assistant can access the current work context as well as—within the scope of a user’s permissions—shared company data.
Copilot is not a single program, nor is it simply “Microsoft’s ChatGPT.” The name refers to a family of products that uses various AI models and services. These include Microsoft 365 Copilot for everyday office work, Copilot Chat, Copilot Studio for creating your own agents, and GitHub Copilot for software development.
Microsoft 365 Copilot combines large language models, Microsoft 365 apps, and Microsoft Graph. When a user enters a prompt—that is, a request—Copilot enriches the request with available context. This could include, for example, content from an open document, emails, calendar appointments, chats, or meetings. An AI model then processes the request and returns a response within the respective application.
The key principle is that of permissions: Copilot is generally only allowed to access information that has already been shared with the logged-in business or school account. The AI does not automatically gain access to the entire tenant. However, permissions that are too broad become easier for Copilot to identify and thus gain greater significance.
This common statement is too vague. ChatGPT is a product of OpenAI, while Microsoft Copilot has its own user interface, orchestration, security mechanisms, and data integration. Microsoft uses large language models from various providers and can choose between models or further develop them. The choice of AI model may vary depending on the function. Therefore, Copilot should not be permanently equated with a single model such as GPT-4.
Furthermore, the quality of a response does not depend solely on the AI model. Other factors that play a role include the prompt, the available sources, the context, data quality, and permissions. Copilot can generate results that are convincingly worded but are nonetheless incorrect or incomplete.
Depending on the product and license, Copilot can incorporate the current work context and approved company information into its responses.
Copilot does not have blanket access to the entire tenant. Existing access rights continue to determine which content a user can access.
Answers, analyses, code, and images may be incorrect or incomplete. Therefore, expert review and clear lines of responsibility remain necessary.
The term "Microsoft Copilot" encompasses several offerings. For businesses, the following versions are particularly relevant.
Microsoft 365 Copilot—already referred to as Microsoft Copilot in some parts of the current Microsoft documentation—enhances the Microsoft 365 apps with work-related AI capabilities and can base its responses on corporate data from Microsoft Graph. This includes, for example, documents, emails, chats, and calendar information that the respective user is authorized to access.
Microsoft Copilot Chat, formerly known as Microsoft 365 Copilot Chat, is available for eligible Microsoft 365 accounts as a web-based chat service with safeguards for corporate data. Without an additional Copilot license, responses are based primarily on information from the Internet and on content that users intentionally provide. The former names “Commercial Data Protection” and “Bing Chat Enterprise” have now been replaced by Copilot Chat with Enterprise Data Protection.
Microsoft Copilot Studio is a low-code platform that enables organizations to develop and manage their own AI assistants or agents and workflows. These can be equipped with instructions, expertise, data sources, and actions, and deployed in Microsoft 365, Teams, or on websites, for example.
Ready-made and custom connectors, programming interfaces, and other tools are available for connecting to business systems. The term “plug-ins,” which was once widely used, is increasingly being replaced in this context by agents, connectors, and tools. Much can be accomplished even without extensive programming knowledge; however, for secure, production-ready solutions, architecture, authorization concepts, testing, and governance remain tasks for experts.
GitHub Copilot is designed to support developers' work. In development environments, the AI assistant can autocomplete code in real time or generate code from natural-language comments and commands. Among other things, it helps with boilerplate code, refactoring, troubleshooting, and debugging. It can also generate explanations, README files, API documentation, and suggestions for unit tests.
The AI can take into account the context of open files or—depending on the mode used—other parts of a project. This gives developers more time to focus on architecture, business logic, and complex decisions. GitHub Copilot can also flag potential errors, security issues, and best practices during code reviews. However, these suggestions are no substitute for automated security scans and tests, nor for human review.
In a lab study published by GitHub, the Copilot group under investigation completed a defined programming task about 55 percent faster. This points to potential productivity gains, but it is not a universal guarantee for every team or project. Copilots are most effective as supportive partners—not as a complete replacement for developers. Generated code must always be reviewed for functionality, security, maintainability, and licensing risks.
Integration with the familiar 365 apps is the key difference from a standalone AI chat. Copilot supports tasks right where documents, data, and communication are created. The exact feature set may vary depending on the license, app, platform, and release version.
In Word, Copilot creates drafts, outlines, and summaries. Existing text can be rewritten, shortened, or adapted to a desired tone. Users can reference relevant files to provide context. This speeds up the process, but does not relieve users of the responsibility to verify factual statements, sources, and wording.
In Excel, Copilot helps you understand spreadsheets, create formulas, clean up data, and identify patterns or outliers. You can ask questions in natural language, such as: “Which product group showed the largest deviation from the budget in the last quarter?” Whether the result is reliable depends on a clear spreadsheet structure and accurate source data.
In PowerPoint, Copilot can create a presentation draft from a description or a suitable source file, consolidate slides, and revise content. In the Create section of the Microsoft Copilot app—still referred to as the Microsoft 365 Copilot app in some parts of the documentation—you can also generate images from text descriptions. This feature is sometimes still searched for under its older name, Image Creator. Any generated image should be checked for factual errors, unwanted details, and appropriate usage rights.
In Outlook, Copilot can summarize long email threads, create complete email drafts, and provide feedback on tone, sentiment, and clarity. Depending on the account, client, and license, it can also prioritize messages and help you prepare for meetings. Certain features are available only for the primary Exchange Online mailbox; therefore, having an email address alone is not sufficient as a technical requirement.
In Teams, Copilot summarizes chats and meetings, identifies discussion points, and can answer questions about the conversation. This helps latecomers get up to speed more quickly and makes it easier to see what tasks have been agreed upon. The results depend on available transcripts, chats, and access permissions. Users should compare the summary with the actual conversation, especially when making decisions.
Copilot can streamline routine tasks in everyday work, speed up the first draft, and consolidate information from a wide range of shared resources. This frees up time for more creative, analytical, or customer-focused work. On PCs and mobile devices, the user experience remains largely within the familiar Microsoft applications; Copilot Chat is also accessible via the Microsoft Copilot app, Teams, Outlook, and Microsoft Edge, among others. However, Copilot isn’t a cure-all and doesn’t automatically improve unsuitable processes or poor-quality data.
However, simply providing a license does not automatically lead to increased productivity. Employees need appropriate use cases, clear guidelines, and practice in working with prompts. Equally important are well-maintained data, transparent filing structures, and a realistic understanding of the limitations of generative AI.
The Microsoft Copilot Dashboard in Viva Insights provides actionable insights into readiness, usage, adoption, and impact. For example, organizations can see how many licensed users are active and in which apps Copilot features are being used. Depending on the license inventory, additional comparisons, agent information, and group-based analyses are available.
The dashboard does not directly measure the quality of each individual task. Metrics such as “Copilot-assisted hours” are estimates based on usage patterns and studies. They should therefore be evaluated in conjunction with qualitative feedback, process metrics, and specific business objectives.
Copilot may produce inaccurate results, misjudge context, or provide outdated information. Answers, analyses, code, and images therefore require human quality control. The more important a decision is, the more thoroughly sources, calculations, and conclusions should be verified.
Permissions management also deserves special attention. Copilot respects existing access rights; however, rights that have been granted incorrectly or too liberally will remain incorrect. Before implementation, companies should therefore establish binding policies regarding sensitive content, sharing, retention, data protection, responsibilities, and permissible use. A robust IT governance framework connects people, processes, and technology.
According to Microsoft, prompts, responses, and data retrieved via Microsoft Graph in Microsoft 365 Copilot and Copilot Chat are not used to train the underlying base models. Nevertheless, organizations should review what data users are permitted to enter, how web search and agents are configured, and which additional services have their own privacy policies.
Microsoft 365 Copilot is generally an add-on product for businesses that already have a suitable base license. For customers with Microsoft 365 Business Basic, Business Standard, or Business Premium, Microsoft currently offers Microsoft 365 Copilot Business as an add-on. For larger organizations, Microsoft 365 Copilot is also offered as an add-on license to a qualifying Microsoft 365 plan. Since Microsoft may change product names, requirements, bundles, and prices, the specific licensing situation should be verified before purchase.
"Copilot Pro" no longer appears as a separate standard plan in Microsoft's current Swiss consumer overview. Instead, Microsoft 365 Single, Family, and Premium are listed there with different AI features and usage limits. Older articles should therefore not present "Copilot Pro" as a standard plan that is still available for purchase without first verifying the current information.
Microsoft 365 Copilot supports German, English, and numerous other languages. The previous statement that it supported nine languages is outdated; in February 2026, Microsoft listed more than 40 language variants. However, individual Copilot features may have different language or regional limitations.
Precise prompts usually improve the quality of the results. A useful prompt includes four elements: the desired goal, the relevant context, specific requirements, and the expected output format.
For example, instead of “Write an email,” it’s clearer to say: “Draft a professional email to existing customers. Summarize the three changes from the attached document, specify October 1 as the deadline, and conclude with a specific call to action. Maximum 150 words.”
Good results often come about in several steps. Review the initial response, provide specific feedback, and have individual sections improved in a targeted manner. Phrases such as “Support this statement,” “Identify areas of uncertainty,” “Use only the sources listed,” or “Shorten the text by one-third” provide additional clarity. Confidential information should only be processed in accordance with applicable company policies and through authorized accounts.
A concise summary of the most important answers regarding licenses, data access, language, devices, and everyday use.
Some Copilot features are included in personal Microsoft offerings or with eligible business accounts. Microsoft 365 Copilot, with comprehensive integration of the work context, is offered to businesses as an add-on license to a qualifying plan. Agents or additional services may also incur usage-based costs. Therefore, it is not just the product name that matters, but the required feature set.
No. Microsoft 365 Copilot accesses data for which the signed-in user already has permission. This prevents blanket, tenant-wide access but does not correct existing permission errors. A thorough readiness check should therefore verify whether sensitive documents, Teams, and SharePoint sites are shared only with the individuals who are actually authorized to access them.
Yes. German and English are among the languages supported by Microsoft. Prompts and responses are also available in a number of other languages. However, an app’s user interface and the language support for a specific Copilot feature are not necessarily the same. For international rollouts, you should therefore verify that all required scenarios are available in the respective languages.
No. Depending on the product and license, Copilot is available in desktop, web, and mobile apps, as well as in browsers. Microsoft Edge provides direct access to Copilot Chat, but it is not the only way to access it. Individual features may vary across Windows, macOS, the web, iOS, and Android, which is why organizations should test their actual device and app environments.
Copilot automates parts of tasks but bears no responsibility for the outcome. People must set goals, evaluate results, and make decisions. This applies to customer-facing content and data analysis just as much as it does to programming: An assistant can increase speed and creativity, but it cannot replace expertise, control, and accountability.
Before a broad rollout, it is recommended to conduct a Copilot Readiness Assessment. It evaluates not only licenses and technical requirements, but also data access, governance, security, data protection, meaningful use cases, and training needs. Following this, a pilot with selected user groups can demonstrate where efficiency gains actually occur and where processes or data need to be improved first.
With SmartAI365, companies gain a comprehensive overview of Copilot use cases, productivity potential, and key strategic decisions. Complementary Copilot workshops and training sessions provide practical prompts, use cases, and guidelines for everyday work.
Microsoft Copilot can boost productivity and creativity because AI features are integrated directly into Microsoft 365 and other work tools. The greatest benefits are realized with clearly defined tasks, high-quality data, and trained users. Without oversight, governance, and appropriate permissions, however, quality and security risks remain.
Companies should therefore not start by acquiring as many licenses as possible, but rather by defining relevant processes, data access, and success criteria. This transforms a general-purpose AI assistant into a controlled support tool with measurable benefits.
Together, we'll assess the technical and organizational situation and develop a realistic implementation plan.
