Case Study: Baker Hughes improves OEE and root-cause analysis with PuppyGraph

Baker Hughes uncovers $71M in RCA losses with PuppyGraph

Baker Hughes, a major energy and industrial equipment company, faced the challenge of diagnosing poor Overall Equipment Efficiency (OEE) across its plants. Their data was fragmented across multiple siloed systems, making it slow and complicated to manually find the root causes of equipment failures. To solve this, they turned to vendor PuppyGraph to implement a zero-ETL knowledge graph solution.

PuppyGraph built a virtual knowledge graph directly on top of Baker Hughes's existing Databricks Delta tables, unifying five datasets without any data movement. On top of this graph, they deployed a team of AI agents that could answer natural language questions by writing and executing graph queries. This solution allowed them to quickly identify high-loss equipment, quantifying issues like USD 71 million in RCA-related losses at one plant and pinpointing compressors and pressure vessels as the primary drivers of downtime, enabling targeted OEE improvement.

Baker Hughes