Get the newsletter
Microsoft · AI / ML · Filed Dec 8, 2025 · USPTO published Oct 1, 2026 · Oct 1 release

Microsoft Patents a System That Reads Your Document and Suggests What's Missing

Microsoft has filed a patent for a system that lets productivity apps automatically analyze what you're working on and suggest additional content, pulling in an AI language model to do the heavy lifting without you having to ask.

A computing device displays a resume, with an application service providing suggestions for improvement based on analysis by an LLM service. Drawing from patent filing US 2026/0300643 A1.
A computing device displays a resume, with an application service providing suggestions for improvement based on analysis by an LLM service.All 8 drawings ↓
Publication number
US 2026/0300643 A1(US20260300643A1)
Applicant
Microsoft Technology Licensing, LLC
Filed · Published
Dec 8, 2025 · Oct 1, 2026
Inventors
Dhirendra Kumar BHUPATI, Mrinal Kumar SHARMA
Status when we published
Waiting for an examiner
Document
20 claims
US classification
704/9
Parent application
Continuation of 18/313,857 (filed May 8, 2023)(18313857)

What Microsoft's AI content-suggestion system actually does

Ever stared at a half-finished document and thought, "what am I even missing here?" Microsoft is patenting a system that answers that question for you, automatically.

The idea is that an app like Word or PowerPoint watches the content you're working on, figures out the surrounding context (what kind of document it is, what you seem to be doing), and then sends both pieces of information to an AI language model. The AI reads everything and sends back a suggestion for content you might want to add.

You'd see that suggestion pop up right inside the app, as a recommendation. No switching tabs, no typing a separate prompt into a chat box. The app handles the back-and-forth with the AI on your behalf.

From the filing · THE ABSTRACT
The application identifies a context relating to the content, wherein the context comprises contextual information by which to evaluate the content. The application generates a prompt for an LLM service which includes the content and the context.

Translation: The software figures out what your document is about and asks an AI to review it in that specific setting.

How the app builds a prompt and gets a suggestion back

The patent describes a pipeline with three main moving parts: the app, a prompt it constructs, and an external LLM service (a large language model, meaning an AI system like the kind that powers ChatGPT or Microsoft Copilot).

First, the app identifies both the content (what you've written or built so far) and the context (metadata about what the content is and how it should be evaluated). Context might include the document type, the section you're in, or rules about what good content looks like in that situation.

The app then assembles a prompt, a structured instruction sent to the LLM service, that asks the AI to evaluate the content and recommend what supplemental material would improve or complete it.

  • The prompt is submitted to the LLM service automatically, without the user initiating a separate AI session.
  • The LLM returns a suggestion for supplemental content.
  • The app displays a recommendation based on that suggestion inside its own user interface.

The patent is broad about what "supplemental content" means, leaving room for text additions, data, images, or other app-specific material, depending on where this is deployed.

What this means for AI assistants in Office and beyond

The problem this addresses is real: people routinely produce incomplete documents, reports, or presentations not because they're careless but because they don't know what they're missing. A system that flags gaps in context, rather than just running a spell-check or a generic AI chat, is a different kind of tool.

For Microsoft, this fits squarely into Microsoft keeps filing on AI-assisted productivity tools in its Office suite. If this makes it into production, the practical effect is that Copilot-style suggestions become more automatic and context-aware rather than something you have to deliberately invoke. Whether that feels helpful or intrusive to users depends entirely on how well the context-detection works.

Microsoft's 16th filing we've tracked since June in our AI assistants that remember you watch builds on earlier work on predicting typed form inputs and keeping chat memory intact.

Source: US 2026/0300643 A1 at the USPTO (PDF). Full patent text and figures are in the official publication.

Get our take in your Top Stories

Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Microsoft patent news breaks.

There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

The drawings · 8 sheets

8 drawing sheets from US 2026/0300643 A1 · click any drawing to enlarge

Computing device, application, and LLM services., drawing sheet 1 from US 2026/0300643 A1
Sheet 1 Computing device, application, and LLM services.
Steps for content evaluation and recommendation., drawing sheet 2 from US 2026/0300643 A1
Sheet 2 Steps for content evaluation and recommendation.
LLM interface processing content and context., drawing sheet 3 from US 2026/0300643 A1
Sheet 3 LLM interface processing content and context.
Content analysis and prompt generation flow., drawing sheet 4 from US 2026/0300643 A1
Sheet 4 Content analysis and prompt generation flow.
Resume analysis and suggested improvements., drawing sheet 5 from US 2026/0300643 A1
Sheet 5 Resume analysis and suggested improvements.
Video search results and recommended tutorials., drawing sheet 6 from US 2026/0300643 A1
Sheet 6 Video search results and recommended tutorials.
Presentation analysis and suggested content., drawing sheet 7 from US 2026/0300643 A1
Sheet 7 Presentation analysis and suggested content.
Computing system with gap analysis process., drawing sheet 8 from US 2026/0300643 A1
Sheet 8 Computing system with gap analysis process.
Patent filing page

Watchlists this filing belongs to

All watchlists →
Reader comments

No comments yet

Start the discussion

Real name or a handle, either is fine. Comments are read by a person before they appear, so allow a little time. Keep it about the filing.