Smart maintenance assistant

Modern motor yachts are equipped with complex propulsion systems and a vast array of technical components. Maintenance and troubleshooting often rely on fragmented, hard-to-access documentation—typically stored as PDFs or printed manuals.

Vessel operators and owners face:

• Inefficient knowledge access (non-digital, non-intuitive formats)
• Long customer support cycles
• High technical entry barriers for non-experts
• Avoidable downtime due to slow issue resolution and unclear diagnostics

This leads to frustration, costs, and underutilized technology—despite high-end equipment on board.

We introduce an LLM-based Multi-Agent System for intelligent monitoring, diagnostics, and predictive maintenance of marine propulsion systems. All agents interact through a secure, scalable orchestration layer that ensures smooth deployment and integration with vessel systems.

At its core is a chat-based assistant embedded in a mobile/web app, enabling users to intuitively access vessel data, documentation, and real-time diagnostics.

Agents

Document Agent (RAG): Retrieves and interprets vessel documentation—including user manuals, service guides, circuit diagrams, spare part lists, construction drawings, and deck plans—using semantic search and vector embeddings.

Equipment Time Series Agent: Provides access to actual and historical measurement data from vessel components (e.g., temperature, pressure, runtime) stored in a time series database for trend analysis and anomaly detection.

Condition Monitoring Agent: Acts as an asset manager that processes SQL-based queries on vessel-specific warnings and alarms. It offers structured explanations, severity classifications, and maintenance recommendations based on alarm patterns.

Orchestrator Agent: Coordinates all agents and manages user interactions through a unified chat interface, intelligently routing queries to the right agent and combining responses into actionable outputs.

Technical Stack Highlights

• LLM backbone with advanced RAG (Retrieval Augmented Generation)
• Embedding models with vector/graph databases
• Modular MAS architecture with clear agent roles
• Production-ready, chat-first UI for end-users and professionals

Source: Alexander Thamm GmbH

For Yacht Owners/Operators: A fun, intuitive tool to understand and manage their yacht like never before – just.

For Technicians: Faster diagnostics, structured data access, and reduced support time.

For Manufacturers: A new digital feature to boost brand experience, customer satisfaction, and differentiation in a competitive market.