Concept: American engineering services company Henderson Engineers (Henderson) has partnered with data infrastructure startup Torch.AI to test a new AI application for designing building systems. Henderson plans to use Torch.AI’s AI solution to analyze complicated engineering and construction documents and diagrams.

Nature of Disruption: The AI solution trains with more than 340 million characteristics using historical construction and engineering data sets, opening up new possibilities for automation and data exchange. The Torch.AI system boasts to read around 1,00,000 engineering drawings with improved accuracy. The startup claims that with an average accuracy of 98.9%, AI can recognize, retrieve, and compare diagrams from project files of more than 40 pages. The software determines what the data is and how it can be beneficial before applying advanced ML to automate the actions required to give new insights and support decisions.

Outlook: The construction industry typically relies on extremely complex information from unstructured data, such as drawings, photographs, portable document formats (PDFs), handwritten notes, and raw text. Businesses are using AI technology to process this data, fundamentally changing operational and analytical capabilities in any information technology (IT) system. The new AI solution helps people reach their full creative and problem-solving potential by reducing the repetitive duties brought on by the expansion of company data and the widening technological barriers. Torch.AI’s solution helps companies in sectors like finance and insurance, building and engineering, and healthcare to simplify building system design. Torch.AI’s clients include the US Department of Homeland Security and the US Department of Defense.

This article was originally published in