Events
Take the Microsoft Learn Challenge
Nov 19, 11 PM - Jan 10, 11 PM
Ignite Edition - Build skills in Microsoft 365 Copilot and earn a digital badge by January 10!
Register nowThis browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
When it comes to deciding on your approach to AI development, there are numerous options to consider, including Microsoft 365 Copilot extensibility options and Azure AI solutions. This article helps you identify the Microsoft 365 extensibility option that is best aligned with your specific needs and goals.
When you're looking to create a fully customized end-to-end AI product to cater to your business needs, you typically choose language models and orchestration for your Retrieval-Augmented Generation (RAG) solutions.
However, with Microsoft 365 Copilot, you also have a choice of extending Copilot's private instances of Large Language Models (LLMs) and the Azure OpenAI service. Copilot's LLM is grounded in each tenant's Microsoft 365 content so it can work with the documents, messages, and other business content people use every day. When you extend Copilot, you provide specific knowledge sources and skills that Copilot's orchestration layer uses to generate responses.
There are two ways you can extend Microsoft 365 Copilot. You can customize Copilot with declarative agents and plugins, and you can add knowledge to Copilot and agents with Microsoft Graph connectors.
Tip
Are you a business decision maker? Check out the Microsoft Copilot Scenario Library.
Declarative agents are designed to enhance the user experience by allowing the creation of personalized chat experiences to provide tailored interactions and responses. Declarative agents have the same look-and-feel as Microsoft 365 Copilot. Declarative agents can be combined with plugins and connectors to add skills and knowledge.
Declarative agents are a great choice for the following cases.
Plugins add skills and actions to declarative agents.
Tip
If you have an existing Teams message extension, you can use it as a plugin with little or no modifications. For new plugins, we recommend creating API plugins.
Microsoft Graph connectors enable data ingestion from various sources to Microsoft Graph, facilitating unified data access and insights across Microsoft 365 and other services. Items ingested via a Microsoft Graph connector are available to Microsoft 365 Copilot as well as declarative agents.
Your journey varies based on your desired outcomes and your coding expertise. Whether you're a seasoned coder or prefer low-code or no-code solutions, there's a suite of tools tailored to your development style.
With Azure AI services, you can create a custom engine agent. Custom engine agents are based on a bring your own model premise. With custom engine agents, your agent is independent of Microsoft 365 Copilot and its LLM and orchestrator.
Custom engine agents are a great choice for the following cases.
How you build and which tools you use are up to you. Your choices are:
The pro-code option allows for full customization. Develop with Teams Toolkit for Visual Studio Code and start with either the Teams AI Library, or Azure OpenAI Studio to configure then export the generated code to Teams Toolkit.
For low-code or no-code, you can develop custom agents rapidly with a user-friendly interface in Copilot Studio.
Note
Microsoft Copilot Studio refers to custom engine agents as custom copilots.
Events
Take the Microsoft Learn Challenge
Nov 19, 11 PM - Jan 10, 11 PM
Ignite Edition - Build skills in Microsoft 365 Copilot and earn a digital badge by January 10!
Register now