← All Case Studies
Mechanical Engineering Industry In-house Solution

AI-powered knowledge management
for machine operators

Embedded AI Interface Edge & Cloud Operations
Demo recording of the knowledge management system in the browser
Demo

Demo recording using an internal test dataset – images from the real customer deployment cannot be published to protect confidential information. This demo was recorded on a laptop – the application runs cross-platform on desktop, tablet, and smartphone.

Background

In manufacturing companies, a large share of operational knowledge resides in the heads of experienced machine operators and maintenance technicians – undocumented, difficult to transfer, and lost when experienced staff leave. Conventional documentation solutions fail in practice: too time-consuming, too far removed from the actual workplace, and too poorly integrated into the workflow.

What we built

A complete, cloud-operated Voice-AI platform for knowledge capture and retrieval – usable directly on the shop floor.

Custom wake-word detection runs in the background on the smartphone – no manual activation required
Certified Bluetooth hearing protection as input device – approved for use in high-noise environments
Voice inputs are automatically transcribed, structured, and persisted in a knowledge base
For queries, the system combines captured experiential knowledge with existing documents (e.g. Word, Excel, PDF)
Custom mobile app as frontend – integrated into the existing workflow
Platform operated as SaaS on European infrastructure, Kubernetes-based
Result

The system is live in production and in daily use. It shows how undocumented knowledge – which would be lost when staff leave – can be systematically captured and kept permanently accessible.

Customer info
Location:
Germany
Industry:
Mechanical Engineering · Paper Processing

Customer and product details are kept confidential at the customer's request.

Tech stack
ASR NLU LLM TTS RAG Cross-Platform App Android Bluetooth Docker Kubernetes Hetzner Cloud
Discuss a similar project →