AI in the Workplace

Using AI to Determine Metadata in SharePoint Document Libraries

Liza

— 3 min read
Using AI to determine metadata in document libraries
Using AI to determine metadata

In this post, I’m going to dive deeper into how I've leveraged AI in SharePoint to help determine and suggest metadata fields for my document libraries. By automating the process of analysing content and recommending structured metadata, I've been able to save valuable time and ensure accuracy in our SharePoint environments. In this post, I’ll walk you through the exact AI prompts I used and the step-by-step process to help you replicate this in your own SharePoint setup.

Please note that wherever I refer to AI I am referring to Copilot and the Copilot Agent that is within SharePoint.

Prerequisites

Before you get started, make sure you have the following:

  • Access to Microsoft Copilot
  • Administrative privileges to modify the document libraries and Term Store.
  • A SharePoint document library with sample content that the AI can analyse.

Step-by-Step Guide to Analysing Your Document Library with AI

Step 1: Set Up the AI Environment

To begin, you’ll need to ensure that your SharePoint environment is set up to use Copilot. Ensure that the SharePoint Copilot Agent is available in your site so that it can analyse document libraries.

Step 2: Gather the Documents for Analysis

Select the SharePoint document library you want to analyse. Copilot will scan documents, looking for recurring terms, patterns in metadata, and keywords in titles and content. It’s best to select a library with a variety of documents that represent the different types of content you store (e.g., policies, forms, procedures).

Step 3: Copilot Analysis Prompt

Here’s the exact AI prompt I used to get started. This prompt will instruct Copilot to analyse the document library and suggest metadata fields based on the content it finds.

Prompt Example 1: General Metadata Analysis

“Analyse the contents of this document library. Identify recurring patterns in titles, keywords, document types, and content. Suggest relevant document types (e.g., Policy, Procedure, Report, Template) and topics (e.g., HR, Finance, Marketing). Additionally, recommend any custom metadata fields that could improve document organisation (e.g., Confidentiality, Approval Status).”

This prompt directs Copilot to scan the library for keywords, document types, and other metadata fields and group documents accordingly.

Step 4: Reviewing AI Suggestions

Once the AI has completed its analysis, it will provide a set of suggestions for metadata fields. For instance, you may receive a report suggesting the following:

  • Document Types: Policy, Template, Procedure, Report
  • Topics: Onboarding, Payroll, Compliance, Training
  • Custom Fields: Confidentiality, Approval Status

At this point, it’s important to review the suggestions to ensure they align with your organisation's needs. You might need to refine or adjust the suggestions to match your existing taxonomy or organisational structure.

Step-by-Step Guide to Refining and Implementing the Suggestions

Step 1: Refining the Suggested Metadata

Based on the AI’s suggestions, take the following steps:

  • Term Store Review: Check your existing Term Store in SharePoint to ensure the suggested metadata fields (like "Document Type" and "Topic") are available. If not, you may need to add them.
  • Consolidate Redundant Suggestions: Sometimes, the AI will suggest overlapping or redundant categories. Group similar terms together (e.g., "HR Policy" and "Onboarding Policy" may fall under the general category "Policy").
  • Validate with Stakeholders: Before finalising, it’s a good idea to run these suggestions by key stakeholders in your organisation to ensure they meet business requirements.

Step 2: Adding Metadata to the Term Store

After refining the suggestions, you’ll want to add them to your SharePoint Term Store for consistency. Here’s how:

  • Navigate to the Term Store: Go to the SharePoint Admin Centre, select the Term Store Management Tool, and choose the appropriate Term Set (e.g., "Document Types" or "Topics").
  • Add New Terms: Based on the AI’s recommendations, click “Create Term” to add new terms (e.g., "Onboarding", "Template", etc.).
  • Publish the Changes: Once you’ve added all the suggested terms, make sure to publish the changes so they are available for tagging in your document libraries.

Step 3: Implementing Metadata in Your Document Library

To apply the new metadata to your document library:

  • Create a New Column: For each new metadata field (e.g., "Document Type", "Topic"), create a new managed metadata column in your document library.
  • Apply to Existing Documents: You can either manually tag existing documents with the appropriate metadata or use Power Automate to automate this tagging based on AI-driven rules.

Additional Tips for Advanced Metadata Management

  • Use Power Automate to Automate Metadata Tagging: Leverage Power Automate to automate the process of tagging documents based on keywords or metadata fields. For instance, if a document title contains the word “Policy”, the flow can automatically tag it as “Document Type: Policy”.
  • AI Training for Specific Needs: You can train AI models in SharePoint Syntex to recognise more specific patterns within your documents. This is especially useful for highly specialised industries like legal or healthcare where documents may have unique structures and requirements.

Conclusion

By incorporating AI to help determine metadata for your SharePoint document libraries, you’re not just saving time — you’re optimising your SharePoint environment for better document management, searchability, and collaboration. Whether you’re working with HR documents or building a general-purpose library, AI takes the guesswork out of metadata creation.

Read more