Data Extraction with Acodis Machine Learning (ML)

Expect 80% faster data extraction with Acodis Machine Learning (ML).

Don't be stuck with template-based solutions that constantly break because of document format changes, opt for a reliable system that works with all of your unique documents.

Data Extraction Robot

Data Extraction Solutions and Challenges

Here's a run-through of typical data extraction processes. Click on the plus to discover the pros and cons of each option. 

Data Extraction in Five Steps

See each step in action by watching our free on-demand demo (no sign-up needed)


1. Upload

Upload any document type to Acodis

2. Classify

Separate documents into different categories

3. Extract

Use machine learning to extract relevant data from the documents

4. Validate

Acodis and/or a human checks if the extracted data is correct

5. Export

Download the extracted data to your preferred location in your desired format

Your Data's in Good Hands

We process and store your data in Switzerland, and if you have dedicated security requirements, we can build a custom IT security architecture for you.

Switzerland | Swiss Data Extraction | Intelligent Document Processing

Acodis Custom Data Extraction Models

Your solution if data extraction accuracy is your top priority.


No Templates, No Code

Your business is unique - and so are your documents. Say goodbye to disappointing tools that don't tackle your data extraction challenges.

Use custom models to get the most accuracy out of your data. Be in charge of how, and where, you want your data to be extracted.

With Acodis, you can build and customise models in just a few clicks - no data scientist needed.

Acodis Pre-Trained Data Extraction Models

A solution that provides you instant results

Fast Results From Day 1

Acodis pre-trained models have been extensively tested and allow you to instantly start extracting data from your documents

Simply drag and drop your most used documents into the pre-trained model and get fast, reliable, results.

Pre-Trained Data Extraction Model | Acodis

Combine Custom and Pre-Trained Models

The real magic happens when you combine our pre-trained models with our custom option to upgrade your data extraction process.


Want to see the magic for yourself?


Machine Learning Made Simple: An In-Depth Webinar


  • What is machine learning?
  • Rule-based vs machine learning systems
  • Supervised vs unsupervised learning
  • How simple it is to build a machine learning model
  • Extract data from PDF documents
  • OCR vs intelligent document processing (IDP)
  • Use Cases
  • Benefits
  • Q&A

Tip Click on the video and hover over the timeframe to select which topic you'd like to watch.

Our customers love Acodis ML

Don't just take our word for it, read what our customers have said.


When it comes to data extraction, Acodis is doubtlessly the right place to go

"Acodis' ML technology enabled us to surprisingly accurately extract elements of complex insurance policies. This will enable us to increase the quality of our customers' coverage and enhance the efficiency of our internal processes in the future."


 Executive Sponsor in Financial Services

Read the full review on


FAQ's: Acodis Data Extraction

Below are some of the frequently asked questions. 
If you want to learn more about building an ML engine, we recommend the webinar linked above.

Couldn't find the answer you were looking for? Take a look at the Knowledge Center!

What do I need to get started with machine learning?

You only need a digitised version of a document you'd like to process, whether it be a scanned copy, PDF, we can even process emails or Excel spreadsheets. 

What languages do you support?

Anything that is read from top left to bottom right. If you would like to process any other language, contact us.

How can you train a data extraction model from Acodis?

Firstly, we need to know what data points you are interested in processing - then we have a look at some example data/documents. Then we have two options to choose from:

1. There is historical data which you can profit from - meaning that we have a PDF and a structured output - and then we reverse-label them, and thus, automatically create a training set.

2. Someone is required to annotate the documents, which can be achieved very easily via Acodis.
You only need to know how to use a computer mouse for you to be able to train a set with us.

Can Acodis ML process handwritten documents?

As of today, we can only extract limited information from handwritten documents. If you know of a technology that is able to do so on a broad scale, please let us know 😊.

How is my sensitive data stored?

We do not store any customer data for longer than it takes to process the document.

Where does the service run?

It is easiest to use Acodis in the cloud.

If you have special requirements for an on-premise solution, the service can be installed on your own machines via a docker container.

What is the difference between a "Document Type" and a "Document Class"?

Document types typically have different data schemes (e.g. purchase orders, policy insurance or pension certificate). In contrast, document classes have the same data scheme but differ among classes in terms of the document's structure or layout (e.g. purchase order client A vs purchase order client B vs purchase order client C).

What are "Data Points"?

Data Points are individual information items, text fields or images, which get extracted from the document. For example, "Name", "Address, "Contract Number", "Delivery Date" etc.

Can the extracted data be validated automatically?

Yes, the extracted data points can be validated automatically against customer-specific reference data or business rules.

Nothing Replaces a Conversation

We're looking forward to speaking with you.