Auto-Generating Digital Twins Using the C3 AI Suite


One of Europe’s largest manufacturing companies delivers billions of dollars of industrial equipment to customers each year across the globe. The company recognized that asset digital twins, which enable monitoring, diagnosis, and predictive maintenance, would help improve service-level agreements (SLAs) for its installed base. The first step toward digital twins is creating and maintaining digital records of the interrelated components that comprise complex assets, also known as digital bills of materials (BOMs). Because downstream changes to asset configuration are typically not reflected in engineering drawings, maintaining accurate BOMs in systems-of-record is a challenging task.

Historically, the manufacturer has employed several hundred technical specialists to maintain BOMs at an annual cost of more than $100 million. Creating a single digital BOM typically requires months of effort to manually extract information from various unstructured data sources, including engineering and electrical diagrams like piping and instrumentation diagrams (P&IDs) and system components tables. The manufacturer sought a scalable, productized solution to perform this parsing and analysis automatically across all its product lines.


The manufacturer tested potential providers using complex gas turbines (GTs) with more than 10,000 components organized into interrelated subassemblies. These GTs perform mission-critical functions for several industries, including electricity generation for utilities and pump powering for oil and gas companies. The customer required a solution that could generate digital BOMs automatically using engineering design and operational data.

A team of three developers and data scientists built an application for digital BOM generation using machine learning and deep learning pipelines in just four weeks. The application has been augmented with functionality for process simulation and failure prediction and alerting, and has later productized it as a configurable SaaS application: C3 AI Digital Twin.

About the Manufacturer

  • Major industrial manufacturer in Europe
  • $90 billion annual revenue
  • 385,000 employees
  • 6 core business units: Power, Infrastructure, Digital, Mobility, Renewables, Health

Project Highlights

  • 4 weeks from diagram delivery to digital twin application demonstration
  • Created logical models representing customer engineering drawing items
  • Developed and deployed deep learning pipelines to detect objects in P&IDs and executed that pipeline on diagrams provided
  • Demonstrated extensibility of deep learning pipeline to adjacent use cases (i.e., detection of components in an electrical diagram)
  • Trained customer developers and data scientists on the C3 AI Suite capabilities


$20 million
Projected cost savings per year
4 weeks
Required to build the application from scratch
Accuracy achievable using application workflows for automated parsing + specialist review
Classes of symbols identifiable by v1 of the application

Solution Architecture

C3 AI Suite

Enterprise AI for Manufacturing

The C3 AI Suite provides the necessary comprehensive capabilities to build enterprise-scale AI applications 25 times faster than alternative approaches. The C3 AI Suite enables manufacturers to rapidly integrate petabyte-scale data from any/all enterprise systems, operational sources, sensor networks, and external providers to power machine learning models that generate predictive insights to solve previously unsolvable problems.

Many global manufacturers are already using the C3 AI Suite to drive digital transformation efforts, generating results such as: reducing inventory by as much as 35%, lowering waste due to quality defects by over 20%, and generating hundreds of millions of dollars in economic value annually.

C3 AI Suite