AI application for MT Aerospace
MT Aerospace is a leading manufacturer of high-performance components for the aerospace industry. makandra developed an AI application for automated document filling for the company.
The MT Aerospace AG is a leading global aerospace company. With over 500 employees, the company develops lightweight structures made of metal and composite materials.
makandra was responsible for developing an AI-powered application that simplifies the process of populating documents with changing data.
Industry
Technology
Services
Our collaboration with makandra has been more than satisfactory in every respect. The quality and speed of the results speak for themselves.
Key Challenges
For a recurring approval process, employees had to gather numerous pieces of information from various source documents and manually transfer them into a fixed template. A solution had to reduce this effort without replacing the existing template.
- Complex, changing data set
MT Aerospace needed an AI solution that could reliably handle a changing data set. The requirements included flexible templates, a variable layout, and the ability to process different types of input. - Changing requirements
Since the requirements for the application frequently changed over the course of the project, the project had to remain adaptable at all times. Considering this, we supported the client in establishing a predictable requirements management process. - Validating the idea with low risk
MT Aerospace initially wanted to test the underlying idea in a cost-effective manner. Therefore, we started with an AI MVP to validate the concept with minimal effort and subsequently moved on to further development.
Our Solution for MT Aerospace
We developed an AI-based application for MT Aerospace that automates the creation of documents during the acceptance process.
The process consists of five steps.
- Set up template: Upload the Word template with placeholders and define which data sources the respective information in the placeholders should come from.
- Upload documents: The uploaded documents serve as context for the system, from which it extracts the information needed to create the target document. It can process any type of file.
- Map documents: The Vision Language Model combines both inputs and determines the appropriate document as the data source for each placeholder.
- Extract values: The required information is automatically extracted from the documents.
- Fill document: Finally, the values are transferred and automatically entered into the Word document – in the existing layout.
Since the solution is operated entirely on-premises with Ollama, data never leaves MT Aerospace’s secure infrastructure.
At the same time, the need for manual searching and transferring of values in documents is eliminated – the pre-filled templates only need to be reviewed. Depending on the project, the customer reports that this can result in savings in the double-digit percentage range.
Sample visualization
To illustrate this, and for data protection reasons, we show the functionality of the AI application using fictitious sample data.
Our Services for MT Aerospace
