SharePoint Just Got a Librarian. Sort Of.
Some of my clients have never opened SharePoint in a browser. They have SharePoint synced through OneDrive, sitting in File Explorer next to Downloads, and until a few weeks ago that meant none of those files ever got tagged with metadata. Now they do — automatically, on upload, whether the file came in through a browser window or got dropped straight into a synced folder nobody has ever opened as a website.
That's the short version. The longer version is why tagging never happened in the first place, and it isn't neglect. Metadata is what actually separates SharePoint from a shared drive — filtered views, automatic categorization, search that returns the right document instead of forty guesses. But applying that metadata has always meant stopping, opening a properties panel, and typing something in by hand, one file at a time. Almost nobody does that consistently, because the one part of SharePoint that makes tagging easy — the browser interface — is the one part most of my clients never touch. They drag a file into a synced folder, it shows up on everyone else's screen a few seconds later, and that's the entire relationship they've ever had with the platform.
Microsoft has been quietly building a fix for exactly that gap, and it's been renamed twice (very Microsoft) while doing it. It showed up in preview last September as "Knowledge Agent." By this spring it was "AI in SharePoint." By mid-June it was "Copilot in SharePoint," and this time the rollout came opt-out by default for every tenant with a Microsoft 365 Copilot license — meaning a lot of businesses got this turned on without anyone actually deciding to turn it on. If you saw "Knowledge Agent" referenced somewhere last fall and have been wondering what happened to it, this is what happened to it. Microsoft documented the change in Message Center notice MC1311968, which is the kind of place a feature goes to get renamed a third time without most admins ever reading it.
It Runs Whether You Open SharePoint or Not
The question I actually cared about, for the File Explorer crowd specifically, was whether that claim would survive contact with how my clients actually work — whether tagging only fires for files uploaded through the browser, the one interface they never use. I checked. It doesn't matter which door the file comes through. The autofill triggers on files landing in the library regardless of upload path, and a file dragged into a synced OneDrive folder gets read and tagged exactly the same as one dragged into the browser window. That's the detail that actually matters here. A feature that only worked for browser uploads would have helped nobody I work with. This one doesn't require anyone to change how they work — it just starts working underneath them.
Under the hood, it reads the actual content of the document — including OCR on scanned files and PDFs — and uses that to auto-populate metadata columns. It also generates AI-powered library views on its own and can build entire sites, pages, and lists from a plain-language description instead of a template picker. Microsoft calls the guardrails around this "Skills" — org-defined standards for structure and metadata that get applied automatically during creation, not reviewed after the fact.
Strong for New, Weak for What You Already Have
Here's the pattern I've watched across two decades of this platform: Microsoft builds for everyone, which means it finishes for no one. This feature is the pattern again. It's genuinely strong at standing up a brand-new library the way it ought to be built from day one — describe the site or the list you want in plain language, and it builds the structure and the columns to match. It's much weaker at reaching into a library your business has already run for years and making sense of what's there. The AI benefit and your existing column structure don't automatically get along, and whether a rebuild is worth it is a library-by-library judgment call, not a blanket yes. This is exactly the kind of assessment I'd walk through library by library with a client before touching anything — what's already working, what's actually broken, and whether the AI columns are worth the disruption of getting there.
Where the Librarian Stops
Three limits are worth knowing before you assume this solves everything. First, it doesn't retroactively organize an existing library — new AI-populated columns sit alongside your old manual ones, not a conversion of them. Second, initial setup only scans the twenty most recent files in a library; anything older needs a manual reprocessing pass to get tagged at all. Third, editing an existing document doesn't re-trigger tagging — only new uploads get the treatment. And not every column type is supported yet: Person/Group, Location, Image, and Lookup columns are still out of reach.
None of that makes it not worth having on. It makes it worth knowing exactly what you're getting, which is a librarian for what comes in from today forward — not a librarian who's already read everything on the shelves. And this isn't a small-business quirk or an enterprise-only rollout. It affects any business with Copilot seats and an active SharePoint library, regardless of headcount, which means plenty of tenants already have this switched on and haven't noticed yet.
It's one more example of a capability that's already running in a tenant whether or not anyone knew to look for it — which is the whole premise of this series. If this is the first time you're hearing that something in your subscription was doing more than you realized, start with the post that maps out the rest of it.
Questions I'm Already Getting
Does this replace the folder-versus-metadata decision? No — it assumes metadata is the model you're using and automates the part people skip. If your library is still organized as deep folder trees, the folder mistake that breaks SharePoint is still the conversation to have first.
Do I need a separate license to get this? No new purchase, but it does ride on a Microsoft 365 Copilot license — if your business already has those seats, check whether this is active in your libraries right now, since it may already be running without anyone having flipped a switch. We've covered what Copilot actually does elsewhere in this series if you're still deciding whether those seats make sense at all.
Will it clean up the messy library I already have? Not on its own. It tags what comes in going forward and reprocesses the twenty most recent files at setup. Anything older, and anything already buried in a folder structure, still needs a deliberate pass — by a person, not the feature.
The value was always in the metadata. It just took until mid-2026 for tagging it to stop being the thing everyone meant to get around to.
Accurate as of July 2026. Microsoft updates its products and pricing regularly.
J. Scott Clark is the President and CEO of The 365 Collective, Inc., a Microsoft 365 consulting and training firm serving small and mid-sized businesses across healthcare, finance, construction, engineering, publishing, and more.
Most of the tools your team needs are already in your subscription. The question is usually just whether anyone has taken the time to set them up.