A document library with three metadata columns, named correctly, with choice options on two of them, created in under 90 seconds from one voice prompt. The agreements library demo shows the entire flow from empty site to ready-to-upload library.
If your team builds document libraries regularly — for contracts, policies, project deliverables, vendor agreements, marketing assets — this is the fastest path to a properly structured library since SharePoint shipped.
The prompt
The demo uses a single multi-clause voice prompt to create the library and all its metadata columns at once.
"Can you please create a document library called Agreements. Now in this Agreements library we want to have three columns. One for Effective Date — that should be a date column. I want one for the Agreement Status — the status options should be Draft, In Review, and Pending Legal Review. We also want a column called Agreement Type with the options being Customer Agreement, Sales Agreement, and Service Agreement."
What this prompt encodes:
- The library name (Agreements)
- Three column specifications, each with name, type, and choice options where applicable
- The column types inferred from the wording (date, choice, choice)
AI in SharePoint parses the prompt, plans the work, checks the site for an existing library with that name first, and then creates the structure.
What AI in SharePoint does behind the scenes
Three things happen in roughly this order.
Duplicate check. The agent searches the site for an existing library called Agreements. If one exists, it surfaces that in chat and asks the user how to proceed. This is a small but important safety net; manual library creation does not check by default and accidentally creating two libraries with the same purpose is a common pattern in legacy sites.
Library creation. The new Agreements library is created with the standard SharePoint library template. Default columns (Name, Modified, Modified By, file size) appear automatically.
Custom column configuration. The three columns described in the prompt are added:
- Effective Date as a date column
- Agreement Status as a choice column with three values
- Agreement Type as a choice column with three values
All three columns are added to the default view so users see them immediately when they open the library.
The whole sequence takes around 45 seconds end to end.
Verification
In the video, the user opens the new library after creation to verify the structure. The columns are present and configured correctly:
- The Effective Date column shows the date picker when edited
- The Agreement Status column shows a dropdown with Draft, In Review, and Pending Legal Review
- The Agreement Type column shows a dropdown with Customer Agreement, Sales Agreement, and Service Agreement
The user uploads three test documents to the library. The columns are editable in grid view, the choice column dropdowns work as expected, and the date column accepts date values.
No additional configuration is required. The library is ready for the team to use.
Why this matters for document workflows
Three scenarios where this pattern saves the most time.
New team site setup. A team's first set of document libraries used to be a half-day of work: figure out the libraries, define the columns, set the choice options, configure the views. With voice creation, the same setup takes under thirty minutes for most team sites.
Standardised library templates. Teams that need consistent library structures across multiple sites (every project gets a Contracts library, every department gets a Policies library) now have a fast way to replicate the structure verbally rather than copying configurations through PowerShell scripts.
Mid-flight column additions. When a team realises mid-project that they need an extra column on an existing library, the conversational flow handles it. "Add a Review Status choice column to the Agreements library with values Awaiting Review, Reviewed, Returned." No navigation to library settings; the change is made in chat.
When you'd want to be explicit about column types
The inference works for most cases but a few situations benefit from being explicit:
Numbers. "Total contract value" could be a number or text. Saying "as a number column" removes the ambiguity.
Currency. Treat as a number column with the appropriate format applied.
People columns. "Owner" or "Contact" could be text or a Person field. Saying "as a Person column" clarifies.
Multi-line text. Notes, descriptions, comments — by default these get treated as single-line text. Specify "as a multi-line text column" if you want the larger field.
Yes/No. A boolean column needs to be specified explicitly; saying "approved or not" might create a choice column with two options instead.
For routine library structures (file metadata, classification choices, dates, statuses), the inference works without explicit type hints.
Where this fits in the broader pattern
This is part of the What to Produce capability in Copilot in SharePoint. The same voice-driven approach extends to:
- Creating SharePoint lists with voice and importing data
- Working with existing list data conversationally
- Generating Word, Excel, PowerPoint files from chat
- Creating interactive HTML dashboards from SharePoint data
The full breakdown of the three Copilot in SharePoint capabilities (What to Know, How to Act, What to Produce) is in the three capabilities article.
What to do this week
If your team builds document libraries more than once a quarter, two actions.
-
Try this on your next library build. Pick a small library you need anyway. Dictate the structure (library name, columns, choice options) as one prompt. Note where the inference lands correctly and where being explicit would have helped. By the second or third library you build, the prompt pattern becomes muscle memory.
-
Put a SharePoint.md context file on the site first so the library structures inherit your team's naming conventions, metadata standards, and required fields. The template is the fastest way to get there. With it in place, you can say "create a contracts library following our standards" and the agent applies the right column patterns automatically.
For organisations replatforming several team sites onto Copilot in SharePoint patterns, the Copilot in SharePoint Immersive is a one-day session where we build the libraries, lists, SharePoint.md, and first three Skills for one of your most active sites together.
