ContractAI

How to transform legal expert knowledge into an algorithm

Our latest innovation development is a service based on Artificial Intelligence, which transforms legal expert knowledge into an algorithm. The service does not only support the legal department with a first and fast risk assessment of contracts, but it also helps c-level management within approval processes. The solution is currently being launched as a beta version on the Bertelsmann Collaboration Platform (BCP), which is a hyperconnected cloud-based shareconomy to host and provide AI services and software solutions across the Bertelsmann universe (more about the BCP here).

How does the ContractAI service work?

As a first step, contracts can be uploaded directly from your email account or even from your iPhone when using the app. After the upload is completed, the contracts are automatically reviewed with the use of the AI-based algorithms. Using deep neural networks, the AI classifies each paragraph of the contract according to predefined categories. These categories include information about topics such as, e.g., liability, warranty, privacy policy, or change of control. After the contract data is processed, the service returns all relevant details such as liability limits, extracted from each paragraph. In addition, the app allows the user to access the paragraph containing the relevant information directly to verify the findings of the algorithm. The app will be designed to allow for human feedback such that the AI will continuously improve its confidence over time. After the contract is analyzed by the algorithm, indicative analysis results can be accessed directly in the app, and a more detailed version of the result, which includes the specific paragraphs, can be sent to any email account of the user’s choice.

Please note that the product is currently only available for IT contracts. In the future, the Product will support a more diverse set of contract formats, e.g., Data Protection Contracts. If you are curious to learn more or would like to have your own contract format supported, please do not hesitate to cont(r)act us.

The successful realization of this project was based on a joint collaboration between several units within Arvato Systems, including the AI core team, the legal department, (which served as innovators and are the product owner), app developers, UX+design units, cloud architects, and data engineers. The authors of this article, Ricardo R. Gameiro, and Dr. Theresa Stocks work as Data Scientists with an emphasis on artificial intelligence and advanced analytics. Their team develops state-of-the-art machine learning solutions for various industries and have experience in the areas of natural language processing, computer vision, and data forecasting.