How to search your scanned documents

You scanned the document. You know it exists somewhere. But when you try to search for it, nothing comes back. The words are locked inside an image, invisible to any search bar. This is the central frustration of working with scanned documents, and it is the reason most people's digital filing cabinets are only marginally better than the paper ones they replaced.

Fabric solves this by treating every scanned document not as a static image but as a source of readable, searchable text. When you scan a page or upload a PDF, Fabric runs optical character recognition (OCR) to extract the text from the image, then indexes that text so you can search it, ask questions about it, and connect it to everything else you have saved. The result is that a receipt you photographed on your phone six months ago becomes just as findable as a note you typed yesterday.

This guide explains how the process works, what kinds of documents it handles, and how to get the most out of searching your scanned content in Fabric.

What happens when you scan a document

The moment a scanned document enters Fabric, whether through the quick capture tool on your phone or as an uploaded PDF, a sequence of processing steps begins automatically.

First, Fabric's OCR engine analyses the image to identify text. It reads printed text in a wide range of fonts, sizes, and layouts, from crisp laser-printed contracts to faded thermal receipts. It handles columns, tables, headers, footers, and text that runs at odd angles. The extracted text is then stored alongside the original image, so you always have both the visual document and its searchable content.

For handwritten text, Fabric's handwriting scanner goes further. Handwriting is notoriously difficult for traditional OCR because letter shapes vary so much between writers. Fabric uses AI models trained on diverse handwriting styles to read cursive, print, and mixed handwriting with a high degree of accuracy. If you have old letters, handwritten meeting notes, or journal pages, the handwriting scanner can extract their content and make it searchable too. This is worth setting up early if you plan to digitise any amount of handwritten material.

Once the text is extracted, Fabric does not stop at keyword indexing. The search system builds a semantic understanding of the content, which means you can search by meaning rather than by exact wording. If a scanned invoice mentions "plumbing repair" but you search for "pipe fixing costs," Fabric can still surface it. This is a significant step beyond what conventional file search offers, and it is particularly useful for scanned documents where you may not remember the precise phrasing.

How semantic search changes what you can find

Traditional search requires you to guess the exact words that appear in a document. With scanned documents, this is especially limiting. You might remember the gist of a letter but not its exact phrasing. You might search for "rent increase" when the letter says "adjustment to your monthly tenancy charge." A keyword-only search would miss it entirely.

Fabric's semantic search understands the relationships between concepts. It maps your query and your documents into a shared space of meaning, then finds the closest matches regardless of whether the exact words overlap. This works across all your content, but it is particularly valuable for scanned documents because you often have less control over how the original text was worded.

The memory engine takes this further by connecting your scanned documents to related content across your entire collection. When you save a scanned insurance document, Fabric may link it to a related policy renewal email you clipped, a note you wrote about switching providers, or a previous claim form. These connections form automatically, building a web of context that makes every individual document more useful. Your scanned pages do not sit in isolation; they become part of a larger, navigable body of knowledge.

Scanning documents with your phone

The fastest way to get paper documents into Fabric is with your phone's camera. The quick capture feature lets you photograph a document, and Fabric handles the rest: straightening the image, enhancing contrast, running OCR, and indexing the text.

This works well for one-off captures, such as a receipt at a restaurant or a notice pinned to a board, and equally well for batch scanning. If you are working through a stack of old paperwork, you can photograph pages in quick succession. Each one is processed independently and becomes searchable within moments. For a more detailed walkthrough of this process, the guide on how to scan paper documents with your phone covers the practical steps.

The quality of the scan matters for OCR accuracy. Good lighting, a flat surface, and a steady hand all help. That said, Fabric's OCR is designed to be forgiving. It handles slight blurriness, shadows at the edges, and minor skewing without losing much accuracy. You do not need professional scanning equipment to get good results.

Working with different document types

Fabric processes a range of scanned content, not just traditional documents. Photos of whiteboards, screenshots of handwritten notes on a tablet, photographed receipts, scanned contracts, medical records, old family letters, and multi-page PDF scans are all handled.

For multi-page PDFs, Fabric processes each page individually and indexes the full document as a single item. You can search across all pages at once and jump to the specific page that contains your match. This is useful for longer documents such as contracts, manuals, or academic papers where you need to find a specific clause or passage.

If you are working with old family photographs that include handwritten captions, letters, or postcards, the combination of the handwriting scanner and OCR means you can search for the content of those handwritten elements. The guide on digitising old family photos and letters covers this use case in more depth.

Asking questions about your scanned documents

Searching is one thing. But sometimes you do not want to find a document; you want an answer that lives somewhere inside your documents. Fabric's AI assistant lets you ask natural-language questions about your scanned content.

For example, you might ask, "What was the total on my car insurance renewal last year?" or "When does my broadband contract expire?" The assistant searches through your scanned documents, finds the relevant passages, and gives you an answer with a reference to the source. You can verify the answer against the original scan, so you are never relying on the AI's interpretation alone.

This capability extends to external AI tools as well. Through MCP integration, you can access your Fabric content from tools like ChatGPT and Claude. These external assistants can query your scanned documents with scoped, time-limited agent access, meaning they can read what you permit for the duration you specify, and nothing more. If you already work within another AI tool, you do not need to switch contexts to search your scans.

Organising your scanned documents

Search works best when it is supported by some degree of organisation. Fabric provides tags and folders for structuring your scanned documents, but it does not require you to organise everything manually before you can find it.

You might create folders for broad categories such as "Financial," "Medical," or "Household," and use tags for cross-cutting labels such as "Tax 2025" or "Warranty." But because semantic search and the memory engine work across your entire collection regardless of folder structure, you can also take a more relaxed approach. Scan everything, organise loosely, and rely on search to surface what you need. Many people find that a combination works well: light folder structure for the categories they access regularly, and search for everything else.

All scanned documents are stored in Fabric's cloud drive, which means they are available from any device. A document you scanned on your phone at a shop is immediately searchable from your laptop at home. There is no syncing step and no storage limit to worry about.

Security and privacy

Scanned documents often contain sensitive information: financial records, medical documents, identity papers, personal letters. Fabric uses AES-256 encryption and holds CASA Tier 2 certification, which means your data is protected both in transit and at rest. The privacy and security page covers the full details of how your data is handled.

When you use agent access to let an external AI tool query your scanned documents, that access is scoped to specific content and expires after a set time. You control what is shared and for how long, so your scanned documents are never exposed beyond the boundaries you define.

Tips for better scan quality and search results

The accuracy of search depends partly on the quality of the text extraction, which in turn depends on the quality of the scan. A few practical habits make a noticeable difference.

Use good lighting when photographing documents. Natural daylight or a well-lit desk lamp works better than overhead fluorescent lighting, which can cast uneven shadows. Place the document on a flat, contrasting surface, such as a dark document on a light desk. Hold your phone directly above the page rather than at an angle, as perspective distortion makes letters harder for OCR to read.

For handwritten documents, legibility matters more than neatness. Fabric's handwriting scanner can handle most handwriting styles, but very small writing, heavily abbreviated notes, or overlapping text can reduce accuracy. If you are scanning handwritten notes that you know you will want to search later, it helps to write with reasonable spacing and consistent letter size.

When scanning multi-page documents, keep the pages in order. Fabric indexes them sequentially, so your search results will reference the correct page number if the scan order matches the original document.

Going paperless

Searching scanned documents is one part of a larger shift towards going paperless. When every document you receive can be scanned, indexed, and searched within seconds, the incentive to keep paper copies diminishes considerably. The guide on how to go paperless at home walks through the broader process, from choosing what to scan first to establishing a workflow for incoming paper.

One of the most common reasons people start scanning is to find old bills and invoices more quickly. If that describes your situation, the guide on finding an old bill or invoice fast is a practical companion to this page.

Frequently asked questions

What types of scanned documents can Fabric search?

Fabric can search any scanned content that contains text, whether printed or handwritten. This includes photographed paper documents, scanned PDFs, photos of whiteboards, screenshots, receipts, contracts, letters, forms, and medical records. If there is readable text in the image, Fabric's OCR will extract it and make it searchable.

How accurate is the OCR for scanned documents?

Accuracy depends on the quality of the scan and the clarity of the original text. For clean, well-lit scans of printed text, accuracy is very high. For faded, damaged, or low-resolution scans, some characters may be misread, but semantic search can often compensate by matching on meaning rather than exact keywords.

Can Fabric read handwritten text in scanned documents?

Yes. Fabric's handwriting scanner is specifically designed to read handwritten text from scans. It handles cursive, print, and mixed styles across a range of languages. The accuracy is highest with legible handwriting on clean backgrounds, but it performs well even with less-than-perfect source material.

How quickly are scanned documents indexed and searchable?

Processing typically takes a few seconds for a single page. Multi-page PDFs take a little longer depending on the number of pages and the complexity of the content. In most cases, a scanned document becomes searchable within a minute of being uploaded or captured.

Can I search scanned documents from ChatGPT or Claude?

Yes. Through Fabric's MCP integration, you can access your scanned documents from external AI tools such as ChatGPT and Claude. Agent access is scoped and time-limited, so you control what is visible and for how long.

Does Fabric store the original scan or just the extracted text?

Fabric stores both. The original image or PDF is preserved in your cloud drive, and the extracted text is stored alongside it for search and AI queries. You can always view the original scan to verify any information.

How is my scanned data kept secure?

All data in Fabric is protected with AES-256 encryption and the platform holds CASA Tier 2 certification. Scanned documents are encrypted both in transit and at rest. Agent access for external AI tools is scoped and time-limited, ensuring your documents are never exposed beyond the boundaries you set.

What if a scanned document is in a language other than English?

Fabric's OCR supports multiple languages. The text extraction works across Latin, Cyrillic, and many other scripts. Semantic search also works across languages, so you can search in one language and find results from documents written in another, provided the meaning aligns.

Can I search across both scanned and typed documents at the same time?

Yes. Fabric's search treats all content equally. Scanned documents, typed notes, saved web pages, and any other content in your collection are all included in search results. The memory engine also connects scanned documents to related typed content automatically.

Do I need a scanner, or can I use my phone?

Your phone is all you need. Fabric's quick capture feature turns your phone camera into a document scanner. It handles perspective correction, contrast enhancement, and OCR automatically. A dedicated scanner can produce higher-resolution images, but phone scans are more than sufficient for the vast majority of documents.

Related pages

• How to scan paper documents with your phone

• How to go paperless at home

• How to digitise old family photos and letters

• How to find an old bill or invoice fast