Copilot for Microsoft 365 semantic index manipulation

[HadiToori]

Hey there!

I am curious about how Copilot M365 searches within organizations at the moment? 
and How can we manipulate the semantic index at the Grounding stage or use tagging the documents to get the search results of our choices?

Regards

[#2 · Accepted solution]

Hello, Copilot for Microsoft 365's semantic index isn't directly manipulable through tagging or keyword stuffing. Copilot uses a semantic index to understand the meaning behind your searches, not just keywords. It analyzes your data (emails, documents, etc.) to create connections and relationships between information.
During the "grounding" stage, Copilot refines your search prompt based on context. It considers the application you're using (Word, Excel, etc.) and adds relevant details from the semantic index to improve accuracy. The semantic index only shows information you have access to according to your organization's security settings. This prevents unauthorized access to sensitive data.
While tagging isn't directly used, clear and concise language in your documents can influence how Copilot understands them. Also, Frame your searches as questions you'd ask a colleague. Copilot leverages natural language processing to understand your intent.

Please click Mark as Best Response & Like if my post helped you to solve your issue, this will also help others to find relevant information.

For more information, you can visit my copilot blog www.copilotwithme.com which is works in-progress and help build a reliable resource for all things in copilot.

Thanks
www.copilotwithme.com

[#3]

That's great insightful knowledge to have regarding Copilot M365.
Btw can we use Copilot plugins or RAG(Retrieval-Augmented Generation) or any machine learning techniques to make Copilot M365 such that its search results are the documents with certain metadata or format or date shows at top?
Regards
Regards