decor
decor

Defining the business process

Start from an existing use case

Start from an existing use case

Everything starts with your use case, what do you want to automate and achieve? Depending on your use case, you can use one of our own pre-built formats or workflows.

Design your own use case

Create your own use case in Studio

Studio is the term we use to allow you to self-service and set up new types of documents we’ve never processed before.

Building blocks

The workflows make use of our building blocks and depending on the use case you will need one or a combination of them.

Extracting information from all types of documents, emails, invoices, contracts
  • Classification

    Before starting to extract information from the document, our machine learning technology is able to derive the document type & language, route it to the correct email inbox and recognize the level of priority in which it needs to be handled all from the context of the document.

    Classification
    Classification of document data
  • Capture

    We eliminate manual data entry as much as possible. Every use case requires specific pieces of information to be extracted. We structure and make sense of the unstructured data and provide the customer with the requested output. We are able to read out logo’s, tables, images, barcodes etc. We go beyond OCR and RPA, being able to work with subtleties for each specific use case.

    Capture
    Capture of document data
  • Validation

    Apply business logic to the extracted and classified data. We can validate if certain conditions have been met; eg.: is the document signed? Are all the necessary documents present? Is the document still valid, are the right boxes checked?

    Validation
    Validation of processes and business rules

Machine learning engine

We help you unlock the potential of state-of-the-art machine learning by providing intuitive access to our highly performant Machine Learning (ML) pipeline. Furthermore, we continuously push the boundaries through dedicated research in collaboration with our academic partner KULeuven.

Machine Learning Process
  • Initial Label Set Creation

    Label set creation: stop blaming the data

    Create high quality data sets quickly with minimal effort by leveraging our intuitive annotation tools. Increase the quality by zooming in effortlessly on suspect data points and by quickly identifying potential ambiguity and inconsistencies between annotators.

  • Training Leveraging Feature Engineering

    Feature engineering: our secret sauce

    We have invested heavily in dedicated feature engineering for the specific business challenges we address. Dedicated textual feature engineering for emails and documents, in addition to visual feature engineering to detect signatures, handwriting, checked boxes etc.

  • Machine Learning Evaluation Stage

    Evaluation: the numbers tell the tale

    You should settle for nothing less than fully traceable metrics. Starting with the end-to-end business value that is realized (e.g., time saved), moving on to the overall performance of the ML pipeline and drilling down into individual predictions to identify potential issues.

  • Machine Learning Feedback Loop

    Feedback: The breakfast of champions

    We believe that creating a high quality revolving reference data should be a part of normal business operations. Whenever we ask for human intervention, be it for quality control or due to low confidence levels, we create an additional data point for training.

Automation

Machines are not up to a 100% confident yet, so we made it possible for a human to review the extracted fields, make corrections if necessary and submit. Our technology will remember these corrections and apply them for future processing.

You are in full control of the level of automation of our product. Meaning you decide whether or not a human should be in the loop of the review process. 

Intelligent Automation Solution

How our product is used

There are 2 main ways of working with our product. Either you use our API keys to immediately connect our technology with yours or you use our complete End-to-End product.

decor
decor
  • Contract.fit API Keys

    API only

    What can you do with the API? With our intuitive API and developer portal you can easily integrate Contract.fit into your own apps/systems.

  • End to End Solution

    End-to-End solution

    Our end-to-end product provides your operators with an easy, user-friendly way to manually review the processed documents. They will be able to manually review the documents in just a matter of seconds to make sure the correct data was recognized, to make corrections and to retrain the machine learning model.

  • Bring your own model icon

    Bring your own model

    Give your AI engineers the tooling they deserve to easily create labelled training data, visualise performance, drill down to get insights on where to focus on next. Furthermore, business users will be able to more quickly benefit from their work by leveraging our integrated workflow.

Want to know more about Contract.fit?

Let us show you around.