How Daimler Trucks North America Built a Living Knowledge Graph of Its Business

Why It Matters

The initiative transforms raw data into actionable knowledge, reducing operational risk and accelerating decision‑making across the enterprise. It showcases a replicable model for other legacy‑heavy organizations seeking real‑time insight and AI‑driven automation.

Key Takeaways

• Graph reveals hidden dependencies across DTNA systems
• Neo4j Aura powers live, cloud‑native knowledge graph
• Claude LLM enables natural‑language queries on network data
• Discovered undocumented jobs prevented split‑over failures
• Future agents may auto‑fix issues without human intervention

Pulse Analysis

When Daimler Trucks North America spun off from its parent in 2021, the company faced a massive data‑integration challenge. Legacy documentation was quickly becoming obsolete, and the risk of missing a critical dependency during the split threatened operational continuity. To address this, DTNA turned to graph technology, selecting Neo4j’s Aura service as a living model of its entire IT landscape. By mapping applications, processes, and network flows as interconnected nodes, the firm could visualize relationships that traditional relational databases hide, turning raw data into actionable knowledge.

The implementation hinges on an automated ETL pipeline that ingests network‑flow records from ExtraHop every five seconds, feeding roughly 2,400 records into Neo4j in real time. These edges and vertices are then exposed through a natural‑language interface powered by Anthropic’s Claude model, augmented with the Model Context Protocol to pull in finance, HR, and supply‑chain datasets. This hybrid approach surfaced “unknown unknowns” such as undocumented SMTP batch jobs and hidden SSH scripts, allowing DTNA teams to remediate risks before the final cut‑over and dramatically reduce reliance on siloed experts.

Looking ahead, O’Shea envisions agent‑orchestrated AI that not only diagnoses issues but also proposes and executes fixes, turning the graph into an autonomous operations hub. Such capability would democratize complex system insights across finance, HR, and field operations, cutting decision cycles from weeks to minutes. For the broader enterprise market, DTNA’s success demonstrates that combining graph databases with large language models can deliver real‑time, cross‑domain knowledge, a blueprint that rivals are likely to emulate as they modernize legacy IT estates.