"If you can’t explain it simply, you don’t understand it well enough."

Albert Einstein, Physicist

THE THEORY

Natural Language Processing

Our expertise lies in making sense of data with the help of artificial intelligence, machine, and deep learning methods. We analyze existing client data and do not only give precise predictions but can also highlight unknown potentials. Our application areas a very broad and diverse as we do offer individual and tailor-made solutions for each client: From scanning the relevant information from massive amounts of text, image, and video data, predicting when to refill a machine, product recommendations to all sorts of forecasts like sales forecast, times of arrivals are possible with historical and current data.

USE CASE

Extracting information from documents

A big industrial manufacturing company takes regularly part in large tendering procedures. These procedures are time-consuming and resource intensive.

digitalisierung challenge
Challenge
These public tenders can even have hundreds of pages containing a lot of information, while for the client only a small amount of this information is needed in order to take part in the tender. The process of finding that information is not only long but also a very one-sided task.
digitalisierung solution
Solution
With the help of labelled datasets for such tenders and machine learning, a system could be developed to give very fast insights as to which information in the tender is relevant for the company that serves as a basis for the offer they are going to make.
digitalisierung result
Result
This system saves the companies not only hundreds of hours over the upcoming months and years, but the process is also less prone to error reducing the one-sided tasks for the employees.

THE STORY

The process

In a kick-off workshop, the client and craftworks were together defining the goals and which information in the tenders are crucial to the client. In a first step, historical tenders were labeled in order to develop a self-learning system and train it. In a next step, both employees and the machine learning algorithm were checking the tender documents in order to compare the results and further train the system to work in a next step fully automatic.

Our clients