Author Byline: Written by the SharePoint Team at Betatest Solutions helping enterprises plan, migrate, and manage SharePoint environments. 

What Is Document Processing for Microsoft 365? The New Home of SharePoint Premium’s AI Features

Document Processing for Microsoft 365 is Microsoft’s content AI service that automatically classifies, tags, extracts, and generates information from documents inside SharePoint and Microsoft 365. Configure it once on a document library, and every file that lands there gets processed metadata filled in, invoice data extracted, summaries written — with no human touching it. It’s billed pay-as-you-go through an Azure subscription, with no per-seat license required.

If you’ve heard the names Microsoft Syntex or SharePoint Premium, this is the same technology those brands were retired and the AI document services now live under this name. This guide covers what the service does, what it costs, and where it delivers ROI.

Key Takeaways

• Content AI on autopilot. Document Processing works in the background on document libraries — classification, extraction, tagging, and generation happen automatically as files arrive.
• Pay-as-you-go pricing. Billed per transaction through Azure. Nearly free to pilot; variable at scale, so estimate document volumes before broad rollout.
• Autofill columns and extraction models are the standout features — AI-written metadata and automated data capture from invoices, contracts, and forms.
• It’s the successor to Syntex and SharePoint Premium. Same capability family, new packaging; governance tools (Backup, Archive, Advanced Management) are now separate products.
• It’s also Copilot preparation. Well-tagged, well-structured content is what makes Copilot accurate — this is how you get there at scale.

Document Processing vs Copilot: Two Different Kinds of AI

The fastest way to understand the product is by contrast with the AI everyone knows:

• Copilot is interactive — a user asks, it answers.
• Document Processing is automation — nobody asks anything. Rules and prompts configured on a library run against every document, at any volume, around the clock.

They’re complementary, not competing. Microsoft explicitly positions Document Processing as how you prepare content for Copilot: Copilot’s answers are only as good as the structure, metadata, and permissions of the content underneath it. We covered the interactive side in our guide to building an intranet with SharePoint and Copilot — Document Processing is the engine room beneath it.

The Core Features

Autofill Columns — AI Metadata Without Manual Tagging

Add a column to a SharePoint document library and give it a natural-language prompt: “extract the contract end date,” “summarize this document in two lines,” “identify the vendor name.” Generative AI answers the prompt for every file — existing and newly uploaded — and writes the answer into the column.

Why this is the flagship feature: metadata is what makes SharePoint searchable, filterable, and automatable, and manual tagging is the thing users never do. Autofill solves document management’s oldest adoption problem. If you run metadata-driven libraries — the approach in our guide to setting up a SharePoint document library without folders — autofill is what populates those columns at scale.

Document Processing Models — Structured Extraction

Train a model on a document type — invoices, purchase orders, application forms — and it extracts key-value pairs and table data (invoice numbers, amounts, dates, line items) into library columns automatically. Models handle structured, semi-structured, and freeform documents.

The classic pairing: extracted data triggers Power Automate workflows. An invoice arrives by email, data is extracted, an approval flow routes it, the record posts to finance — zero manual entry end to end.

Content Assembly — Documents Generated, Not Just Read

The reverse direction: build a Word template with placeholders, connect it to a SharePoint list, and generate standardized documents — contracts, offer letters, certificates, onboarding packs — one at a time or in bulk. Consistency and compliance come built in, because every document starts from approved language.

OCR — Making Scanned Content Visible

Optical character recognition makes scanned documents and images searchable and processable. For organizations sitting on archives of scanned paper, OCR is the gateway service — nothing else can work on content that search can’t see.

Translation and eSignature

• Translation converts files and Stream video transcripts across dozens of languages, automatically and at scale.
• eSignature collects signatures while documents stay inside Microsoft 365 — a compliance and audit advantage over routing files through external signature tools.

Where Syntex and SharePoint Premium Fit In

The product has worn four names since 2019, which is why documentation and vendor advice often conflict:

• Project Cortex (2019): Microsoft’s original knowledge-and-AI initiative.
• SharePoint Syntex → Microsoft Syntex (2020–2023): the first productized version of content AI.
• SharePoint Premium (2023): Syntex rebranded and expanded at Ignite 2023.
• Document Processing for Microsoft 365 (current): the AI document services under their present name — while the governance tools that briefly shared the Premium umbrella became standalone products.

What this means for you: anything you read under the older names describes this capability family — but licensing follows the current structure. If a proposal or article quotes Syntex or Premium SKUs, ask for it to be restated in current terms.

The Governance Products (Related, but Separate)

Three tools formerly marketed alongside these services are now individually priced products:

• SharePoint Advanced Management (SAM): restricted access controls, site lifecycle policies, sprawl reporting, conditional access for sites. Per-user add-on (around $3/user/month; included in Microsoft 365 E5).
• Microsoft 365 Archive: low-cost cold storage for inactive SharePoint content that remains searchable and compliant.
• Microsoft 365 Backup: fast, native backup and restore for SharePoint, OneDrive, and Exchange.

Budget for these separately — and remember tooling supports governance rather than replacing it. The operating model — policies, ownership, review cycles — is covered in our SharePoint governance framework for enterprises.

Document Processing for Microsoft 365 Pricing: How Pay-As-You-Go Works

Unusually for Microsoft licensing, the AI document services are consumption-billed:

1. Connect an Azure subscription in the Microsoft 365 admin center — that’s where charges land.
2. Pay per transaction — typically per page or per document processed, with each service metered separately. As a reference point, autofill columns have been priced around $0.05 per transaction, charged once per document no matter how many autofill columns run on it.
3. No per-seat license, no commitment. Most Microsoft 365 plans qualify, and months with no usage cost nothing.

What this means for budgeting:

• Piloting is nearly free. A few hundred documents costs pocket change — there is no cheaper way to test enterprise AI.
• Scale needs a model. Costs move with volume, so estimate monthly document counts and average pages per document before broad rollout.
• Backlogs are a one-time spike. Running AI over a 10,000-document archive is a single large charge — plan it deliberately rather than discovering it on the Azure bill.

Rates and packaging evolve, so verify against Microsoft’s pay-as-you-go documentation before committing numbers to a budget.

Use Cases: Where Document Processing Automation Pays Off

Finance — invoice and receipt processing. The highest-ROI entry point for most organizations. Extraction models capture invoice data, workflows route approvals, and manual data entry disappears.

Legal and procurement — contract intelligence. Autofill columns extract parties, renewal dates, and obligations across entire contract libraries; date-driven alerts stop auto-renewals slipping through. Content assembly generates new agreements from approved templates.

HR — onboarding at scale. Personalized offer letters and onboarding packs generated from a SharePoint list in one action; incoming employee paperwork auto-classified and filed.

Regulated and records-heavy industries. OCR plus AI classification turns scanned archives into searchable, retention-labeled content — the strongest fit is in financial services, healthcare, and manufacturing, where document volume meets regulatory pressure. Retention connects directly into Purview, covered in our SharePoint compliance and Purview migration guide.

Copilot readiness — the defining 2026 use case. Copilot surfaces whatever content and permissions allow. Before rollout, content needs to be accurately tagged, classified, and governed — Document Processing does that at a scale no manual effort can match. If Copilot is on your roadmap, this is groundwork, not garnish.

Should You Adopt It? A Quick Decision Frame

Strong fit if:

• You process high volumes of similar documents — invoices, contracts, forms
• Scanned archives are invisible to search
• Copilot is planned and content hygiene isn’t ready
• Metadata-driven libraries keep failing because nobody tags manually

Weak fit if:

• Document volumes are low and varied — per-transaction AI needs repetition to pay off
• Basic SharePoint hygiene (structure, permissions, governance) isn’t in place yet — fix the foundation first

The low-risk path: one library, one document type, one autofill column. Pay-as-you-go makes the pilot nearly free, and results are visible within days.

FAQs