SP Editor adds an AI Assistant demo
Software Development Redmond, Washington
Microsoft showed an AI Assistant in SP Editor that brings SharePoint context into Copilot-based help for code, queries, and errors.
Key insights
- Microsoft demoed the AI Assistant in SP Editor during a community call.
- The assistant works inside the browser extension for SharePoint developers and power users.
- It can use SharePoint context, PnPjs, Graph SDK, search, and REST request data.
- Generated code can be applied, and in some cases applied and executed immediately.
- The demo uses GitHub Copilot CLI and a local bridge rather than a Microsoft-hosted model.
Microsoft recently published a demo video titled Introducing the AI Assistant in SP Editor showing a new assistant inside SP Editor, the browser extension for SharePoint developers and power users. The presentation was given by Tomi Tavela as part of a Microsoft 365 and Power Platform community call. The main takeaway is simple: SP Editor can now pass live SharePoint-related context into an AI assistant so developers can inspect, explain, and act on what they are already working on.
The video is not a product announcement with rollout details. It is a walkthrough of how the assistant fits into an existing developer workflow, and it shows the tool alongside SP Editor features such as SharePoint context inspection, PnPjs and Graph SDK consoles, web and list properties, REST calls, search queries, file editing, customizers, site templates, theme colors, and web part property copying. Microsoft also points viewers to the SP Editor repository and browser extensions for Edge and Chrome.
What's new
The new part of the demo is the AI Assistant inside SP Editor. According to the material, it is powered through GitHub Copilot CLI and a local bridge. That means the assistant is presented as part of the editor experience rather than as a separate web app that users have to copy and paste into.
The assistant is shown working with the context of the active feature in SP Editor. The source material says it can receive PnPjs code, Microsoft Graph SDK code, search queries and results, and SP Shooter request and response data. The demo also shows conversation history and a model picker for choosing among the Copilot models available in the local setup.
Why it matters
This is most relevant to SharePoint developers and power users who already use SP Editor for inspection and troubleshooting. Instead of moving between the browser extension, a documentation tab, and a separate AI tool, they can stay in the same workflow while asking questions about the current request, query, or code.
The value is practical rather than flashy. If you are debugging a REST call, checking a search result, or comparing Graph SDK behavior, the assistant can work with the exact context that is already on screen. The notes also say generated code can be applied directly to the editor or applied and executed immediately, which shortens the loop between trying an idea and testing it.
For admins and makers who support SharePoint solutions, the benefit is faster troubleshooting. The material specifically calls out interpreting REST responses, permission errors, search syntax, Graph SDK exceptions, and unfamiliar API behavior. The demo suggests the assistant can reduce manual context gathering, especially when the problem sits inside a single editor session.
How it works
The source material describes the assistant as a small chain of components rather than a standalone model. SP Editor provides the SharePoint-focused environment, context integration collects the relevant code or data from the active feature, GitHub Copilot CLI provides the local AI interaction layer, and the selected Copilot model produces the response. The result is then used back in the editor through apply or apply-and-run actions.
The material shows several ways to feed information into the assistant:
- Use the active PnPjs, Graph SDK, or SP Shooter context.
- Send search queries and results into the prompt.
- Add a search-result property by using the Robot icon shown in the demo notes.
- Pass console errors or warnings from PnPjs or Graph SDK operations for explanation.
The assistant also keeps a conversation history, and users can move through earlier prompts with the up and down keys. That is useful when refining a question or comparing output after changing the underlying SharePoint context.
Limitations and open questions
The video is a demo, so it does not give rollout timing, licensing, tenant eligibility, or regional availability. It also does not state whether the feature is generally available or preview-only. We can say that Microsoft showed the assistant in the context of a community call, but the material does not tell us how broadly it is being released.
There are also technical details that remain unclear from the material. For example, the source does not explain whether the local bridge requires any specific setup, what the model picker includes by default, or how data is handled beyond the statement that the assistant is powered through GitHub Copilot CLI and a local bridge. The demo shows capability, not deployment guidance.
The source set is also thin. It gives us a clear feature summary and workflow examples, but not deep documentation or customer scenarios. For that reason, readers should treat this as a guided look at the idea and the interface, not a complete product reference.
Sources
Frequently asked questions
What is SP Editor in this demo?
The material describes SP Editor as a browser extension for SharePoint developers and power users. The demo shows it alongside inspection tools, consoles, search, REST calls, and editing features.
What does the AI Assistant use as context?
The notes say it can use PnPjs code, Microsoft Graph SDK code, search queries and results, and SP Shooter request and response data. It can also take console errors and warnings.
Can the assistant run code for you?
Yes, the notes say suggested code can be applied directly to the editor or applied and executed immediately. The video does not give more detail about safeguards or confirmations.
Is this generally available?
The material does not say. It is presented as a demo during a community call, but there is no rollout status, licensing note, or availability statement.