Skip to content

Agentic AI

AI, Smart Glasses, and Knowledge Graphs for the Future of Intralogistics

How a manufacturing industry leader rethought intralogistics by combining AI, agents, and semantics to control AMR fleets and warehouse data in real time.

Industry
Manufacturing, Logistics

AI, Agents, and Semantics for the Intralogistics of the Future

Modern intralogistics requires governing heterogeneous systems, distributed data, and autonomous robots from multiple vendors. This project aims to make the entire warehouse searchable and controllable through natural language by combining conversational AI, Knowledge Graphs, and specialized agents.

The client

A highly-specialized manufacturing company, leader in its sector, operating complex logistics processes and extensive AMR fleets across warehouses and production lines.

The challenge

The company needed to rethink how it managed AMR fleets and logistics data:

  • Robots from different models and vendors, each requiring different control systems
  • Fragmented warehouse data accessible only on a point by point basis
  • A mix of real time data and complex, unstructured information such as manuals
  • Lack of a unified, queryable view of the logistics ecosystem

The solution

BitBang developed an advanced solution combining Smart Glasses, LLMs, Knowledge Graphs, and AI Agents.

The solution enables:

  • Robot selection via Smart Glasses and voice commands regardless of ambient noise
  • Automatic translation of natural language commands into AMR specific control languages using LLMs
  • A Knowledge Graph modeling warehouse structure, data distribution, and real time observations
  • Specialized data repositories: VectorDB for unstructured content and SQL databases for tabular and near real-time data
  • Dedicated agents for specific tasks, ensuring efficiency, contextual control, and reliability

The results

  • Full warehouse querying through natural language
  • Immediate access to current status and historical data
  • Support for Monitoring, Anomaly Detection, Predictive Maintenance and What if Analysis
  • Reduced hallucination risk through guided LLM usage and semantic context
  • More robust management of near real time data and operational context

Let's talk

Facing a similar challenge? Tell us about it: that's where we start.

More Case Studies