Microsoft has officially confirmed that Copilot Chat will soon be able to search and retrieve text from scanned PDFs and images embedded in documents, a long-awaited capability slated to arrive in July 2026. The revelation came via an update to the Microsoft 365 Roadmap on June 30, 2026, where item 559613 was tagged as in development. This upgrade promises to close a critical gap in Microsoft’s AI-powered search experience, bringing optical character recognition (OCR) directly into the Copilot Chat interface for the first time.

For countless professionals and enterprise users, the inability to search inside scanned documents or imaged-based files has been a glaring omission. Until now, Copilot Chat could only index and retrieve information from native digital text in Microsoft 365 files—Word documents, emails, Teams chats, and the like. Scanned PDFs, faxes, and screenshots remained invisible to the AI assistant. The updated roadmap entry marks a pivotal shift, indicating that Microsoft is weaving advanced OCR and computer vision models into its Copilot stack to make every byte of textual information discoverable.

Roadmap Details and What’s Changing

The roadmap entry, identified as feature ID 559613, states: “Copilot Chat will be able to search for and match text found inside scanned PDFs and images embedded within documents in Microsoft 365. This includes text in JPEG, PNG, and TIFF images, as well as scanned pages within PDF files.” The feature is listed as rolling out to General Availability in July 2026, targeting all Microsoft 365 Copilot licensed users.

Currently, Copilot Chat—accessible via Microsoft Teams, Outlook, and the Microsoft 365 web app—relies on the underlying Microsoft Graph to index content. Graph has traditionally lacked native OCR support, meaning that rasterized text was simply ignored. The new capability leverages Azure AI’s document intelligence services, specifically the prebuilt OCR and layout analysis models, to extract text from non-searchable files during indexing. This means that once the feature is live, users can type natural language queries like “find the invoice from Acme Corp sent last March” and Copilot will surface not only email attachments but also scanned PDF invoices that previously would not have appeared.

The Technology Behind the Scenes

At its core, this update represents a marriage of Microsoft’s Azure Cognitive Services with the Copilot orchestration layer. Azure’s Form Recognizer and Computer Vision APIs have long been able to extract text with high accuracy from images and scanned documents. Now, those capabilities are being integrated behind the scenes as part of the Microsoft Graph indexing pipeline. When a user uploads or receives a scanned file, the system will automatically run OCR, store the extracted text as metadata, and make it available to Copilot’s retrieval-augmented generation (RAG) pattern.

One technical nuance is that the OCR will occur at the time of ingestion, not at query time. This ensures responsiveness—users won’t experience delays while a document is being processed on the fly. Microsoft has hinted that the feature will support over 70 languages for text extraction, with plans to expand as the models evolve. Importantly, the process respects existing Microsoft 365 compliance boundaries; extracted text is encrypted and adheres to the same data residency and security policies as the original file.

The practical implications are broad. Law firms, for example, often deal with massive troves of scanned court filings and legacy documents. Paralegals and attorneys using Copilot Chat will now be able to instantly locate case-relevant text without manual OCR preprocessing. Similarly, healthcare organizations can search patient records that were digitized as image-only PDFs, while academic researchers can query scanned historical manuscripts stored in SharePoint and OneDrive.

Accessibility stands to gain significantly. Sight-impaired users who rely on screen readers will benefit from the underlying OCR layer, as extracted text can be surfaced not just in search results but also in document previews within Microsoft 365 apps. This aligns with Microsoft’s broader commitment to inclusive design.

Small and medium businesses will also find relief. Many SMBs still receive invoices, receipts, and contracts as scanned attachments. Copilot Chat’s new ability to “see” into those files means that expense reconciliation, contract review, and compliance audits can become conversational tasks. A user could ask, “What was the total amount on the last three utility bills?” and Copilot will parse the scanned PDFs to deliver an answer.

Integration with Windows and the Microsoft 365 Ecosystem

On Windows, the feature will be most apparent in the Microsoft 365 Copilot app, which aggregates content across the user’s entire digital workspace. Because Windows 11 deeply integrates OneDrive, SharePoint, and Teams, the OCR-powered search will span local files that are synced to the cloud. Even scanned documents stored in a desktop folder will be searchable if the folder is backed by OneDrive.

Microsoft has also indicated that the same OCR technology will eventually power cross-application scenarios. For instance, if a user receives a scanned image in a Teams chat, they can forward it to Copilot Chat with a question like “extract the address from this image” and get an instant response. While the initial rollout focuses on chat-based search, the roadmap suggests that Copilot in Word and Excel will later be able to use the extracted text for summarization and data extraction tasks.

Potential Challenges and User Concerns

Despite the promise, several questions linger. The accuracy of OCR on low-resolution or handwritten documents has historically been variable. Microsoft claims that Azure AI’s latest models achieve character-level accuracy above 95% on clean text, but heavily degraded scans may still produce errors. Users may need to verify extracted information, especially in high-stakes contexts.

Privacy-conscious enterprises may worry about the processing of sensitive scanned content. Microsoft has clarified that OCR processing happens within the customer’s tenant boundary and that extracted text is not used to train foundational models. However, some industries may require additional assurances or even the ability to disable OCR at a granular level—something not yet detailed in the roadmap.

Another concern is storage overhead. Extracted text metadata will increase the index size, potentially affecting SharePoint storage quotas. Microsoft has not yet disclosed if there will be any impact on storage costs, but historically, OCR text additions have been treated as part of the existing file overhead without additional charges.

Rollout Timeline and Availability

The roadmap specifies July 2026 for General Availability, but the usual caveats apply. Microsoft often rolls out features in waves, starting with Targeted Release tenants, then extending to Standard Release over several weeks. Users can track progress via the Message Center in the Microsoft 365 admin portal. There is no indication that the feature will be limited to specific licensing tiers beyond the standard Microsoft 365 Copilot license, which is required for access to Copilot Chat.

Insiders who are part of the Microsoft 365 Copilot Early Access Program may get a preview in the spring of 2026. If past feature rollouts are any guide, some capabilities—like OCR of images embedded in PowerPoint files—might lag slightly behind the initial PDF focus.

How It Compares to Competitors

Google Workspace has offered OCR-powered search via its Search Cloud and Google Vault for years, though its integration with Gemini AI is still maturing. Salesforce’s Einstein OCR provides similar capabilities within the CRM context, while standalone tools like Adobe Acrobat Pro have long offered OCR search within individual PDF files. What sets Microsoft’s approach apart is the seamless integration into the Copilot chat experience—no need to open a separate application or manually run OCR. It’s always-on, background processing that feeds directly into the AI assistant’s knowledge base.

Apple’s Spotlight on macOS has included live text extraction from images for local files since macOS Ventura, but it lacks the enterprise-grade cloud cross-device indexing that Microsoft 365 provides. Microsoft’s implementation promises to unify results across devices and colleagues, making it a true collaborative tool.

Expert Analysis: A Productivity Game-Changer?

Analysts who track the AI productivity space have long pointed to unstructured data as the final frontier for enterprise search. Forrester Research has estimated that over 80% of organizational data is unstructured, and a significant portion of that is images and scanned documents. By enabling Copilot Chat to index this dark data, Microsoft could significantly widen the utility gap between standard Microsoft 365 offerings and those with Copilot.

“This isn’t just a niche update for archivists,” said one unnamed Microsoft MVP who has been briefed on the feature. “It’s about making AI truly aware of all your content, regardless of format. For knowledge workers who deal with external communications, it’s a lifesaver.” The MVP also noted that the feature could reduce the need for third-party OCR add-ins, potentially saving organizations thousands in licensing fees.

However, some remain cautious. The success of the feature will hinge on real-world accuracy and the ability to handle the messy reality of scanned documents—skewed angles, coffee stains, and multi-column layouts. Microsoft’s Azure AI has been battle-tested in products like Syntex and Document Intelligence, so the foundation is robust, but the Copilot integration is a new frontier.

IT administrators looking to leverage the new feature should start auditing their SharePoint and OneDrive libraries for scanned content. Ensuring that files are stored in a modern document library with proper metadata already applied will maximize the value of the OCR indexing. Microsoft recommends that organizations review their compliance and data security policies to account for the newly indexable text, especially if some scanned documents contain sensitive information that was previously considered “hidden” because it was not searchable.

Additionally, training users to formulate natural language queries will help them get the most out of the upgrade. Copilot Chat supports conversational follow-ups, so once a scanned document is found, users can ask for summaries, translations, or specific data points.

The Broader Vision: A Unified Information Assistant

This roadmap entry is part of a larger strategy to make Copilot the ultimate front-end for all organizational knowledge. In recent months, Microsoft has added grounding in web data, enterprise connectors to third-party apps, and on-device Copilot features for Windows 11. Searching inside scanned documents fills a major piece of that puzzle. Looking ahead, Microsoft has hinted at bringing similar OCR capabilities to Copilot for Security and Copilot for Sales, where image-based threat intelligence or business cards could become conversational resources.

The update also dovetails with the upcoming Microsoft 365 Copilot “Recall” feature for Windows, which captures a timeline of user activity. If a user viewed a scanned document, Recall might automatically make it searchable via OCR, creating a seamless bridge between local and cloud-based recall.

Final Thoughts

When the July 2026 update lands, Microsoft 365 Copilot users will find their AI assistant noticeably smarter about the documents that matter. The ability to peer inside scanned PDFs and embedded images transforms Copilot Chat from a text-centric search tool into a multimodal knowledge worker. For organizations drowning in unsearchable legacy files, the feature could immediately become the most compelling reason to adopt Copilot.

Microsoft’s commitment to making every byte of information accessible and secure is clear, but the real test will come when millions of users begin flooding the system with decades of scanned archives. If the OCR holds up, Copilot Chat may just become the enterprise search standard for the next decade.