The Challenge
Imagine having to oversee thousands of documents detailing complex legal obligations or having to maintain various infrastructure elements. How do you efficiently process and disseminate this critical information? Especially when some of these contracts date back to the mid-20th century and are handwritten in legal jargon and technical language, making them particularly difficult for contractors and third parties to navigate. This is what our customer struggled with.
Recognizing the need for a more streamlined approach to managing this overwhelming amount of data, the organization sought a solution that could automate the extraction of key information from these documents to improve accessibility and understanding for all stakeholders.
The approach
Could generative AI be a solution to their challenges? The organization was keen to find out and reached out to Eraneos to help them out. After developing several potential use cases, Eraneos experts started work on a Large Language Model (LLM) solution that aimed to address some of the documentation challenges head-on.
The project kicked off with a development phase during which we looked at the available documentation and crafted tailored prompts for the model. This allowed us to identify and extract crucial information relating to the maintenance, legal obligations, and contractual detail documents. Through iterative testing and refinement, we demonstrated the model’s potential to significantly reduce manual effort while increasing the accuracy and how fast the information could be retrieved.
The customer is responsible for maintaining railway infrastructure, allocating rail capacity, and controlling rail traffic. This includes ensuring the tracks and signals are in good condition, managing the scheduling and usage of train routes, and ensuring the safe and efficient movement of trains.
The result
A proof-of-concept demo was developed and showcased to the customer, successfully demonstrating the solution’s efficacy. Testing confirmed that the solution could streamline the management of the company’s vast documentation, potentially setting a new benchmark for the sector. The client was pleased with the results and expressed confidence in implementing this solution in the future. Initially, it is expected to reduce the workload and accelerate processes, eventually taking over most of the time-consuming labor with minimal human oversight.
By automating the extraction of key information from legal documents, we have introduced a level of efficiency that couldn’t be obtained before. The Generative AI solution is able to not only simplify access to critical data for the customer and its contractors, but to also ensure that maintenance and legal obligations are clearly understood and managed more easily.
This approach also opens the door to further innovation, such as integrating a chat function for querying the system, which would make the search for specific information even more straightforward. The AI-driven assistant is able to provide instant access to essential information while reducing the dependency on manual document reviewing.
Let’s create sustainable change together.
Philipp Pytel
Managing Director – Analytics
Data & AI
philipp.pytel@eraneos.com +49 (0)40 809 081 – 172