AI glossary · Integration
OCR
In one sentence
OCR (optical character recognition) is the technology that turns the text in an image, a scanned PDF or a photo into digital text you can edit, search and process. With AI, it also understands the document's layout and structure.
Also known as: optical character recognition · text recognition · document scanning to text · intelligent document processing
What is OCR?
To a computer, a scanned PDF or a photo of an invoice is just pixels. It sees a picture, not letters. OCR fixes that: it finds the text in the image and turns it into characters you can copy, search or load into a database.
Put it to work
UTIO's AI, through an API
From €29/mo + VAT
For developers and agencies building their own products.
See the AI APIThe technology has been around for decades, but AI has taken it up a level. Traditional OCR recognised one letter at a time and got lost in tables, columns or crumpled paper. Today’s models, based on computer vision and sometimes on multimodal models, read the whole document and understand its structure: what’s a header, what’s a table of line items and which figure is the total.
Related to this term
AI API
- Transcription, voice, text, image, OCR and embeddings
- Per-client keys with a spending cap
- Prepaid wallet: €50 minimum top-up
From €29/mo + VAT Panel account, plus usage
See the AI APIHow it works
- Prepare the image. Straighten, crop, boost contrast and remove shadows.
- Detect. Find the blocks of text, lines, tables and boxes.
- Recognise. Turn each area into text. Neural models use context: an unclear character in the amounts column is probably a digit, not a letter.
- Structure. Return an organised result, for example JSON with supplier, date, net amount, VAT and total, ready to pass to another system through an API.
Just as speech to text turns voice into text, OCR turns images into text. Both are the way in for a language model to work with information that didn’t arrive as typed text.
Example
A drinks wholesaler in Sheffield receives dozens of signed delivery notes every day, which drivers photograph on their phones. Someone in the office used to type in the customer, date and quantities delivered. Now an OCR system reads each photo, extracts those fields and matches them against the order. Only notes with discrepancies or an illegible signature go to review. It’s a textbook case of AI automation: AI reads, rules decide and a person handles the exceptions.
OCR at UTIO
UTIO’s AI API includes OCR: scanned invoices, delivery notes and forms turned into structured text. It uses the same key as the API’s other capabilities, such as transcription, text to speech, text, image and embeddings, so you can chain steps with a single provider: read a document, summarise it and save the result to UTIO’s CRM.
Each key is created in the panel with its own spending cap and rate limit, which helps when you process documents in batches. The full reference for parameters and formats is in the panel.
Frequently asked questions
What is the difference between OCR and document data extraction?
Classic OCR returns the text exactly as it appears. Data extraction goes a step further: it understands that “Total” goes with 1,250.00 and that the date at the top is the issue date, then returns tidy fields. Many tools now combine both.
Can OCR read handwriting?
Current systems read neat handwriting fairly well, but make more mistakes than with print. For handwritten forms, check the key fields or ask people to write in capitals.
What image quality does it need?
The better the image, the better the result: good light, a flat document, no shadows and enough resolution for small print to stay legible when zoomed. A skewed phone photo often works, but with more errors in figures and small print.
Related terms
- Computer vision
Computer vision is the branch of AI that lets a system interpret images and video: recognising objects, reading text, spotting defects or describing a scene. Today it mostly relies on neural networks trained on large numbers of images.
- API
An API (application programming interface) is a set of rules that lets one program ask another for data or an action: you send a request in an agreed format and get a response back, without needing to know how the other system works inside.
- AI automation
AI automation means using artificial intelligence models to handle tasks that used to need human judgement, such as reading a message, working out what it asks for and choosing the next step, rather than just repeating fixed steps defined in advance.
- Multimodal AI
A multimodal model is an AI model that can take in or produce several types of data, such as text, images, audio or video, and connect them. For example, it can read a photo of an invoice and answer questions about it.
Updated on