# Use-Cases - Document AI - AI Powered OCR - GUC > How CloudFiles' AI-powered OCR reads scanned pages, handwritten forms, and images inside Salesforce, then extracts the exact fields or line items you need. *Document AI for Salesforce · OCR & Extraction* ## AI-powered OCR for Salesforce CloudFiles reads scanned pages, handwriting, and images inside Salesforce and turns them into structured, usable data. No separate OCR tool, no manual retyping. [Book a Demo](https://meet.cloudfiles.io/salesforce?utm_content=sf-document-ai-use-cases-ai-powered-ocr-hero) *Loved and trusted by leading companies* Logos: General Mills, Dun & Bradstreet, Wingstop, PenFed, Aviva, Ohio University, CA Conservation Corps, NC State Auditor, Arthrex, Commvault, Computacenter, Aggreko, ITW, Lummus, Northumbrian Water, Hamptons, Anteriad, BrightNight *What AI-powered OCR does* ## From a scanned page to a Salesforce field OCR reads the characters on a page. CloudFiles takes that a step further: it reads the page, then extracts the exact field or line item you asked for, without anyone retyping it. ### Without OCR in Salesforce - A scanned or handwritten document sits on the record as an image. Nobody can search it or pull a field from it without opening it. - Someone reads the page and retypes the name, date, or ID number into Salesforce by hand. - A batch of scans means repeating that by hand for every file in the batch. ### With CloudFiles OCR - The page is read the moment it lands, scanned, photographed, or handwritten, printed or not. - The specific field or line item you asked for comes back as JSON, CSV, plain text, or a table, matched to what your process expects. - It runs against files already in Salesforce or in a connected drive, so nothing has to leave your org to get read. *Where teams use it* ## Top document processing use cases How teams already run OCR and extraction through Salesforce documents. - **Simple property extraction from KYC documents** — Pull a specific field (name, address, ID number) off a single identity document. - **Mass property extraction from medical intake forms** — Run the same extraction across a batch of forms instead of one at a time. - **Mass line-item extraction from tables** — Read the rows and columns off an invoice or statement, not just the header fields. - **Text detection from scanned documents** — Turn an image-only scan into searchable text. - **Text detection in images** — Read printed or handwritten text off a photo, a business card, a whiteboard shot. - **Structured output in the format you need** — Get the extracted data back as JSON, CSV, plain text, or a table, matched to what the next step in your process expects. *How this fits with the rest of Document AI* ## OCR is the first read. Here's what happens next. - **[Document Intelligence](/salesforce/document-ai/use-cases/document-intelligence)** — Once OCR has turned a page into text, Document Intelligence is what reasons over it: summarizing a long report, matching two differently-worded agreements, scoring a response. - **[Splitting & Comparison](/salesforce/document-ai/use-cases/splitting-and-comparison)** — A single upload is often several documents merged into one PDF. Splitting separates them first, so OCR and extraction run against each document individually instead of one long, mixed file. - **[Custom Document Approval Workflows](/salesforce/document-ai/use-cases/custom-document-approval-workflows)** — The fields OCR pulls out (a name, a date, an ID number) are what an approval checklist checks against. Missing or mismatched data is what gets a document routed back before a human ever has to spot it manually. *Why teams pick CloudFiles for OCR* ## Built to fit how Salesforce documents actually work - **Reads files already in Salesforce** — CloudFiles reads a file that's already in Salesforce directly, not only files sitting in an external drive, so you don't have to move a document out of Salesforce first to read it. - **Native to Salesforce** — OCR runs inside Salesforce itself. There's no separate app to open and no file to export before it gets read. - **Output matches your process** — Extracted data comes back as JSON, CSV, plain text, or a table, whichever your Salesforce flow or downstream system already expects. - **Your files, your storage** — Documents stay in your own storage. CloudFiles never re-hosts them, and they are not used to train CloudFiles' AI models. ## How CloudFiles answers questions about a document Ask a document a question in plain English, or another language, and get a precise, structured answer back instead of a highlighted snippet. For example, "What is the expiry date of this driver's license?" returns "01/08/2029." A query can also reference a value already on the Salesforce record as it runs. [Read the help doc](https://help.cloudfiles.io/salesforce/intelligent-document-queries) ## Stay compliant & secure. CloudFiles is independently audited for SOC 2 Type II, certified to ISO 27001, and compliant with HIPAA and GDPR. Data residency is supported in the US, EU, UK, and AU. Your files stay in your own storage, CloudFiles never re-hosts them, and are not used to train our AI models. Certifications: SOC 2 (Type II), HIPAA (Compliant), ISO 27001 (Certified), GDPR (Compliant) ## Your customers won't wait. Book a demo and see CloudFiles read one of your own scanned or handwritten documents live. [Book a Demo](https://meet.cloudfiles.io/salesforce?utm_content=sf-document-ai-use-cases-ai-powered-ocr-footer) ## AI-powered OCR, answered directly **Q: What is AI-powered OCR in CloudFiles Document AI?** It's the part of Document AI that converts scanned pages, photos, and handwritten forms into text and structured fields, directly inside Salesforce, so a document that arrives as an image can still be searched, queried, and extracted from. **Q: Can CloudFiles read handwriting, or only typed text?** Both. Handwritten forms are one of the document types OCR is built to read, alongside typed and scanned text. **Q: Is OCR the same thing as data extraction?** No. OCR turns an image into readable text; extraction is the separate step that pulls a specific field, like an expiry date or an ID number, out of that text. CloudFiles does both, but they're two distinct operations under the hood. **Q: What formats can the extracted data come back in?** JSON, CSV, plain text, or a table, whichever matches what your Salesforce flow or downstream system expects. **Q: Does OCR work on files stored outside Salesforce?** Yes. It runs against files in Salesforce as well as files in a connected external drive. The document doesn't need to be copied into Salesforce first. **Q: Does CloudFiles use my documents to train its AI models?** No. Your files stay in your own storage and are not used to train CloudFiles' models. See the compliance section above.