Maschinenfabrik Reinhausen (MR), a global leader in transformer control and regulation technology, leverages an AI-driven solution developed by appliedAI to automate the processing of complex specification documents - making the configuration and quoting of MR products more efficient, scalable, and consistently high in quality.
Read on to see how MR develops AI-based solutions across the end-to-end workflow - from chatbot-assisted specification extraction to automated product matching - accelerating processes, ensuring quality, and enabling sustainable scalability.
Timeline from Nov 24 to Nov 25 for the project described in 2 seperate phases
Initial Situation and Problem Statement
The project by Maschinenfabrik Reinhausen (MR) and appliedAI addresses low process efficiency and a high susceptibility to errors in the handling of specification documents. Today, key information must be manually extracted from specifications with varying structures—an time-consuming step that is prone to mistakes and often results in inconsistent outcomes. In addition, the quality of how documents are interpreted and applied varies significantly depending on individual expertise, experience, and regional working practices.
By automating specification extraction and subsequent product matching, the project aims to noticeably reduce processing time and costs at MR while sustainably increasing and standardizing overall output quality.
Approach and Methodology
The project is delivered by a specialized team of AI engineers with many years of experience in deployment and productization, working closely with MR’s domain experts. This tight collaboration results in a solution that is both powerful and user-friendly, directly addressing MR’s needs in specification processing: an intuitive, AI-powered chatbot is combined with a robust, fully automated backend pipeline and seamlessly transitioned into a production-ready end-to-end workflow.
The core solution consists of an AI-powered chatbot with an intuitive interface that enables Q&A on specification documents, as well as a fully automated pipeline that extracts relevant specification details and supports customer-specific output formats. In addition, the solution matches the identified requirements against selected MR products, creating a direct link between documentation and the product portfolio. To ensure long-term reliability and scalability, the core solution is embedded in a secure, enterprise-grade architecture that remains fully under MR’s control.
All of this is planned and delivered within a clear project structure that not only provides the solution but also enables MR to continuously evolve it.
The initiative is set up as a six-month project in two phases: the first phase focuses on design and prototyping, and the second on hardening and transitioning into production operations. Across both phases, a strong emphasis is placed on enabling MR to run the solution independently over the long term, improve it in a targeted manner, and scale it as needed.
Outcome and Next Steps
Building on the joint project with appliedAI, MR is able to use AI to translate complex specifications into actionable product requirements within minutes - accelerating and scaling product configuration and quote generation processes while simultaneously stabilizing and ensuring quality.
Results:
- Increased processing speed
- ~45x for QA ➔ LLM average Q&A response: ~3 s (vs. 134 s for manual
response) - ~1500x for Automated Extraction ➔ Specification extraction (70-pages
document): ~12 seconds (vs. ~5 h manually, for asic analysis and detailed analysis) - ~830x for Automated Product Matching ➔ Product matching (70-
pages document, 3 products): ~13 s (vs. estimated 3 h manually) - High extraction accuracy
- 90% for basic analysis extraction results
- 77% for detailed analysis extraction results
- ~45x for QA ➔ LLM average Q&A response: ~3 s (vs. 134 s for manual
- Fast and Scalable Adoption: Solution is production-ready and actively used by domain experts.
Next Steps:
Following the successful implementation, the next phase focuses on evolving the solution into a fully scalable, enterprise-ready platform. It is designed to be driven by MR’s domain experts while being seamlessly embedded into MR’s core tool landscape. A no-code, editable knowledge base enables non-technical users to maintain products and independently configure extraction and matching pipelines. A central configuration database closes the feedback loop by feeding function-specific domain knowledge directly into the pipeline’s continuous refinement.
In parallel, deep integration with product management and quoting tools lays the foundation for automatically synchronizing extracted and matched product data with the enterprise product catalog and leveraging LLMs for intelligent quote generation. A scalable integration framework also enables rapid connection of additional internal systems (e.g., ERP, analytics, or CMS) and ensures the solution can evolve sustainably in line with MR’s strategic and operational requirements.