Optimise inventory management: Crateflow enables precise AI-based demand forecasts

A key challenge for companies is to control overstocking and understocking and to make supply chains resistant to disruption. Demand forecasts that make it possible to precisely plan factors such as stock levels, order quantities, production capacities and procurement strategies can help with this. The founding team "Crateflow" at RPTU is developing an AI-supported software platform for this purpose. The project is being funded by an EXIST start-up grant from the Federal Ministry for Economic Affairs and Climate Protection and the European Social Fund. Crateflow will be presenting its prototype at the Hannover Messe from 22 to 26 April at the Rhineland-Palatinate research stand (Hall 2, Stand C36).

Markus Heidt uses a case study to explain how companies benefit from precise demand forecasts: "Imagine you need a bigger car because your family is growing. You research the brand you trust, find the right vehicle and then find out from the manufacturer that the delivery time is over 12 months. This is annoying and leads you to buy the car from another manufacturer who can deliver faster. From the manufacturer's point of view, this means that a customer is lost because production cannot keep pace with the development of demand."

The two founders want to close this gap with their data-driven software and offer it as a support tool for demand and procurement planning. Alongside Markus Heidt, Daniel Antonatus is part of the start-up project - they both know each other from their student days. With Crateflow, they are developing a solution that provides precise demand forecasts, enabling companies to effectively adapt their stock and production strategies. This allows both customer satisfaction and resource efficiency to be maximised.

How does the Crateflow solution work? Prediction models that the founders customise to specific user scenarios serve as the infrastructure. The software initially requires company-related data as input, for example from an ERP (enterprise resource planning) system. It also receives supply chain-relevant information, such as raw material prices, container freight rates or current world events, via external interfaces. Crateflow uses a wide range of AI methods to analyse and link all this data into precise forecasts. "The basic version will still require companies to export data from their ERP system. The long-term vision is that we will create a platform that companies can access directly," says Heidt.

Special technical features of the Crateflow solution: the consideration of external characteristics and the integration of disruptions in real time allow companies to develop proactive supply chain management strategies. Forecasting intervals are also used, which enables supply chain planning experts to better understand the scope and uncertainty of the AI model. At any point in time, it is clear how confident the AI model is in a prediction. Crateflow does not provide a black box, but transparent and explainable data.

At the Hannover Messe, the founders will present their development to date to interested companies and demonstrate the benefits of Crateflow based on initial results. "We are looking forward to the dialogue - especially when it comes to possible use cases and therefore requirements for our software," says Antonatus.