Industrial companies are moving beyond experimental AI projects toward systems that can deliver measurable results across plants and operations. In a Cognite interview, Graham Upton, head of technology and innovation at Capgemini Engineering UK, explains why trusted, contextualized data is becoming central to that transition.
Capgemini focuses on combining engineering, operational technology, data, AI, and business transformation. Its work spans asset-intensive sectors including manufacturing, energy and utilities, chemicals, aerospace and defense, transportation, and life sciences.
Upton argues that successful industrial AI requires more than sophisticated algorithms. Companies need reliable data, domain expertise, governance, and the ability to integrate AI into daily operations. Capgemini combines these capabilities with Cognite Data Fusion, which provides a contextualized industrial data foundation.
The partnership has helped organizations in manufacturing, energy, and life sciences build data environments intended to improve decision-making, asset performance, and operational risk management. One important lesson is that companies increasingly want to scale proven AI applications across multiple facilities rather than continue running disconnected pilot projects.
Industrial knowledge graphs and data contextualization are becoming particularly important. Many organizations already collect large volumes of information, but data without operational context can be difficult for AI systems to use effectively. Connecting information through digital threads and knowledge graphs can make it more useful for predictive, autonomous, and agentic applications.
Upton recommends beginning AI transformation with a specific business problem instead of choosing technology first. Engineering, operations, IT, data teams, and business stakeholders also need to collaborate closely.
Looking ahead, Capgemini expects industrial AI to progress beyond dashboards and analytics toward decision intelligence, industrial copilots, autonomous operations, and agentic systems. Trusted industrial data, the partners argue, will provide the foundation needed to deploy these technologies responsibly and at enterprise scale.
