MS Copilot expands with MS Graph connectors

Editor's Notes

• #2: Many of you have been working with Microsoft 365, its apps and services for a while now. Inside the Microsoft 365 ecosystem, Microsoft Search, SharePoint Search and now a days, Microsoft Copilot helps you to surface your relevant search results from the people you work with, data or information you have access to from various apps including content and files from SharePoint and OneDrive, conversations from Microsoft Teams, and emails from Outlook. External data dependency adds another angle to this equation. On a daily basis, alongside Microsoft 365, we interact with various external systems for e.g. ServiceNow to work on tasks or assigned tickets and incidents, Workday for for financial management and human resources, or even any form of data including any REST API, any database, etc. How about we can bring all this data together and form an analysis on top of it using Microsoft Copilot? Yes, it is possible with Microsoft Graph Connectors and in today’s session we are going to talk about it. Microsoft Graph connectors provide a powerful platform for ingesting unstructured, line-of-business data into Microsoft Graph. This integration allows Copilot for Microsoft 365 to reason over the entirety of your enterprise content.
• #6: SMITA Let’s jump into it: What is Microsoft 365 Copilot and how does it work exactly? Microsoft 365 Copilot is your copilot for work.  Microsoft 365 Copilot is a sophisticated processing and orchestration engine that provides AI-powered productivity capabilities. At a fundamental level, Copilot has a few core components: Large language models (LLMs). Copilot is more than OpenAI’s ChatGPT embedded into Microsoft 365. It’s a sophisticated processing and orchestration engine working behind the scenes to combine the power of LLMs, including GPT-4, with the Microsoft 365 apps and your business data in the Microsoft Graph that is now accessible to everyone through natural language. Content in Microsoft Graph, such as emails, chats, and documents that you have permission to access. The Microsoft 365 apps that you use every day, such as Word and PowerPoint. (Microsoft 365 apps (such as Word, Excel, PowerPoint, Outlook, Teams, loop, and more) operate with Copilot to support users in the context of their work.) Copilot combines the power of large language models with data: your data in the Microsoft Graph and the knowledge of the internet (not just at a point in time when the large language model was trained, but real time, up to date knowledge) . It can be accessed through natural language across all your Microsoft 365 apps - to turn your words into the most powerful productivity tool on the planet. https://learn.microsoft.com/en-us/microsoft-365-copilot/microsoft-365-copilot-overview#copilot-integration-with-graph-and-microsoft-365-apps
• #7: The plugins and connectors that we build for Copilot are part of the larger Copilot for Microsoft 365 system. The diagram on the screen illustrates how a user's prompt is interpreted and processed into natural language results through the Copilot system. Let's summarize how Microsoft 365 Copilot works, starting with the user prompt and ending with the Copilot response: Copilot receives an input prompt from a user in an app, like Word or PowerPoint. Copilot then pre-processes the prompt through an approach called grounding, which improves the specificity of the prompt, ensuring that you get answers that are relevant and actionable to your specific task. It does so, in part, by making a call to Microsoft Graph and Semantic Index and accessing your organization’s data. Semantic Index scopes the user's data used by Copilot to the documents and data that are already visible to them through existing Microsoft 365 role-based access controls. This retrieval of information is referred to as retrieval-augmented generation (RAG). It allows Copilot to provide exactly the right type of information as input to an LLM. When doing so, Copilot combines this user data with other inputs such as information retrieved from knowledge base articles to improve the user's prompt. Copilot takes the response from the LLM and post-processes it. This post-processing includes more grounding calls to Microsoft Graph, responsible AI checks, security, compliance, and privacy reviews, and command generation. Copilot returns a recommended response to the user, and commands back to the apps where a user can review and assess the suggested response. Copilot iteratively processes and orchestrates these sophisticated services to produce results that are relevant to your business because it’s contextually based on your organization’s data.
• #8: Copilot for Microsoft 365 helps people work better with Microsoft programs like Word, Excel, and Outlook. It's good at tasks like understanding, summarizing, predicting, and translating. It uses information from your organization's files and messages to do these tasks. But, most business tasks don't only use Microsoft 365. With Copilot extensibility, you can add extra abilities and knowledge that fit your business. This makes it possible for Copilot to do even more amazing things for your organization. You can make Copilot better suited to your needs by turning your own apps into tools that work with it. Also, you can give Copilot more information from your company by connecting it to your data through Microsoft Graph connectors. X- Axis: Expand Organizational knowledge. Ground Copilot in your organization’s content and data to create even deeper insights and connections. Y-Axis: Extend users’ skills. Connect Copilot to the applications you work with plugins to streamline your everyday tasks and activities. Microsoft 365 Copilot extensibility enables you to expand your organization’s AI-powered productivity, capacity, and creativity. At the same time, it Preserve trust, as Copilot operates within your organization’s security and compliance boundaries and adheres to Microsoft’s responsible AI standards.
• #9: SMITA There are two ways of extending Copilot for Microsoft 365: by augmenting Copilot with skills through plugins, and grounding it with organizational data through Graph connectors. Plugins: Plugins expand your users' skills by interacting with your web service using natural language in Microsoft Copilot. With plugins, you can: Access real-time information such as finding the latest news coverage on a product launch Retrieve relational data such as reporting on service tickets assigned to a given team member Perform actions across apps such as creating a new task in your organization's work tracking system You can build plugins by building a Microsoft Teams message extension or a Power Platform connector. If you already have a message extension or Power Platform connector, then you already have the foundation to create a plugin for Copilot.  Microsoft Graph connectors Graph connectors increase the discoverability and engagement of your enterprise data by deeply integrating your data into the Copilot for Microsoft 365 experience. With Graph connectors, you can: Make the most of your external data by letting Copilot read and summarize data from various places. This helps you understand things better. Use Copilot as a research assitance, It can find information, summarize it, and answer questions using the data you choose. Surface the intelligence of Copilot in Microsoft Search and other tools to improve how users find what they need. Explain diagram Semantic index The semantic index is a system that helps improve the way search functions work within Microsoft 365. It's like a database that organizes information in a smarter way, making it easier for users to find what they're looking for. This system understands the meaning behind words and phrases, so it can provide more relevant search results. Additionally, it respects security rules, ensuring that users only see content they're allowed to access. Overall, the semantic index enhances the search experience within Microsoft 365 by making it more efficient and accurate.
• #10: Now, lets talk about Extensibility options for Microsoft Copilot for Microsoft 365. You can customize Copilot for Microsoft 365 in different ways using external services, apps, and data. If you want to create your own connector or plugins, you can choose the tools and software development kits (SDKs) that best suit your needs.
• #11: SMITA You may want to explore the pros and cons of each option before making a decision. For instance, Graph connectors help keep third-party data within the compliance boundary by indexing the data. On the other hand, the real-time data through the use of message extensions are not indexed and can be shared or accessed outside of the defined compliance boundaries. As a plugin developer, you are responsible for securing your customer's data within the bounds of your service and providing information on your policies regarding users' personal information. READ TABLE
• #12: SMITA Here is an example of writable data with a message extension. This plugin manages a product inventory, which Copilot searches the data from the inventory and displays the results in Adaptive Card, where it allows a user to modify the stock.
• #13: SMITA This screen shows a sample response from Copilot, utilizing data from Graph connector. The Graph connector indexes issues and repositories from GitHub, presenting the results in plain text as opposed to an Adaptive Card format. By comparing this with the Message extension example shown earlier, you can see the differences between them.
• #15: This sample contains a Microsoft Graph connector built in PowerShell that shows how to ingest REST Countries (https://restcountries.com/) content. For each country, it extracts the metadata from the REST API, maps them to the external connection's schema and ingests the content retaining the content and metadata. The ingested content is set to be visible to everyone in the organization. https://github.com/pnp/graph-connectors-samples/tree/main/samples/powershell-json-restcountries # Create Entra app .\Initialize-EntraApp.ps1 # Create external connection .\Initialize-ExternalConnection.ps1 # Import content .\Import-Content.ps1 urlToItemResolvers urlToItemResolver enables the platform to become aware when users share URLs from your external content with each other.
• #16: To ensure your Graph connector works well with Copilot, there are several things you should do. These tips improves relevance for your connector content with Microsoft Search: By including urlToItemResolver in your connector’s activity settings, Copilot will become aware when users share URLs to your content with one another. When users share an item with one another, Copilot boosts its importance for including the item in its response. Provide a rich human-readable name and detailed description when you create the connection. This helps Copilot understand what the data source is, so it includes the data source in its response if it’s relevant. Apply semantic labels. Semantic labels help Copilot understand the semantic meaning of your schema. You must apply as many of them to your schema as applicable. Add user activities. In addition to adding your content externalItems to the Microsoft Graph, your connector should also add user activities around your content. Items that have more activities are boosted in their importance. Open M365 Admin Center Settings > Search & intelligence Open Data Sources tab.
• #18: Information stored outside of Microsoft 365 can be seen by everyone in your organization or just a specific group. The rules about who can access this outside information are kept in the system where it's stored. When you bring in content using Microsoft Graph connectors, you get both the content and its rules from the outside system. With this info, you make a list of who can access each piece of content. Then, when you bring it into Microsoft 365, you include this list. This helps make sure that only the people listed in the outside system can see the content you brought in. When you use Microsoft Graph connectors to bring in outside content, you can choose different ways to make sure the same people who could see it outside can still see it inside Microsoft 365.
• #19: Access Control List is like a list of rules. Each rule has three parts: Access type: It says if someone can or cannot access something. Type: It tells us what kind of person or group the rule is for. It could be someone from your company, a group in your company, everyone in your company, everyone except visitors, or a group from an outside system. Value: It's who the rule applies to, like a person's name or a group's name.
• #20: https://learn.microsoft.com/en-us/graph/use-postman https://devblogs.microsoft.com/microsoft365dev/use-microsoft-graph-connectors-to-securely-bring-external-content-into-microsoft-365/ https://learn.microsoft.com/en-us/graph/connecting-external-content-external-groups