SaaS solution for precise and scalable data extraction_

Orbit is a brand of John & John Industry Solutions GmbH and offers a cloud-based platform that enables companies to automate, monitor and optimize their logistics processes – beginning with route planning, and ending with customer service. The platform offers comprehensive features such as GPS tracking, driver management and digital portals to improve efficiency and transparency in the supply chain.

Project goal and challenge

The aim of this project lay in providing support to logistics companies in efficiently processing large volumes of transport requests. Every day, these companies receive up to 1,000 emails with often unstructured and incomplete information which needs to be processed quickly and accurately to avoid losing potential customers. The challenge was how to structure these queries in an automated way and respond quickly in order to effectively manage the volume of data and ensure customer satisfaction.

Measures and development

In its solution cronn uses Large Language Models (LLM) to extract a structured JSON output from the unstructured input data such as e-mails, PDFs or images, and to make it available via an interface. This in turn enables automated analysis of transport orders. This process involves extracting transport-relevant information such as material dimensions, shipping addresses, and quantities from sources such as goods invoices and e-mail signatures. This can then be supplemented with additional data, such as the type of freight.

The following steps were taken:

• Data Science: Collection and analysis of data sets; Building an ELT data pipeline with AWS S3 as the data lake followed by automatic tidying, convertion, and storage of data as jobs in AWS DynamoDB
• Machine Learning: Selection and evaluation of different AI models (e.g. OpenAI, Mistral, Anthropic); Application of prompt engineering and OCR technologies to extract text and images from documents
• Agile software development: Implementation of the business logic in a robust Java/Spring backend; Development of a frontend for job monitoring with TypeScript/React and integration of the AI components in Python
• Cloud Computing: Building the Cloud Environment with Terraform on AWS; Scaling the AI components in Lambda workers using an asynchronous job pipeline

Customer benefit

The tight-knit cooperation between cronn and Orbit resulted in the development of a cloud-native SaaS solution which both automates and structures e-mail data. The result was not only increased efficiency, but also a learning opportunity for both partners: while Orbit gained in-depth insights into modern AI methods, cronn was able to deepen its know-how in real-world logistics processes.

Simultaneously, professional evaluation and quality assurance plans ensure that the AI models are continuously monitored and that improvements are quickly incorporated into production. This means that data extraction remains precise, scalable and adaptable to new requirements. The result: a versatile tool which relieves the strain on companies, drives innovation and is being constantly optimized through the cooperation between Orbit and cronn.