Larsen & Toubro is shaping its artificial intelligence strategy not merely around internal productivity gains but also as a future go-to-market play where tools developed in-house for own businesses could eventually be offered to clients across industries.
The engineering and infrastructure major’s presence across construction, manufacturing, energy and services gives it an advantage in building AI applications rooted in real operating environments rather than lab-based experimentation, said R Ganesan, senior vice president & head – corporate centre, L&T Construction. Ganesan is responsible for driving digital transformation and artificial intelligence (AI) initiatives across the construction business.
L&T is framing its digital, data and AI push as a single, integrated ecosystem anchored by compute infrastructure and scaled through proprietary solutions. The company is building data centre capacity in partnership with Nvidia to create what it describes as an “AI factory”, enabling clients to move from experimentation to full deployment, while also serving as a commercial offering rather than a purely captive asset.
AI Factory
This infrastructure layer is tied to a broader stack spanning data, platforms and applications, with a strong emphasis on India-specific, multilingual and multimodal datasets. The objective is not just to host compute but to develop proprietary solutions that can be diffused across sectors and populations, positioning AI as both a national-scale capability and a business opportunity.
Ganesan said solutions first deployed in L&T’s own businesses could later be scaled for sectors facing similar challenges. That strategy rests on using L&T’s internal operations as live proving grounds. Giving an example from the construction business, Ganesan said that with 350,000-400,000 workmen engaged across projects on a typical day, labour availability gaps of 10-15% create a strong case for automation, robotics and AI-led productivity tools.
“For example, we process around 50,000 invoices as a company. We have deployed agents which help with optical character recognition (OCR) to scan that and get all the details like the GST number, the details of the cost, etc. and match the required fields to ultimately approve the payment,” Ganesan said, citing another example of a scalable use case the construction major could take to the market. He added that a human in the loop is included in this application to ensure accuracy and oversight.
Scalable Solutions for Industry
In construction, the group has developed computer vision systems that can identify whether workers are wearing helmets, gloves or safety shoes, and send alerts to supervisors when unsafe conditions are detected. These systems can be mounted on cameras, drones or mobile robots.
Ganesan said the company has also identified renewable energy projects, where algorithms are being used to identify factors that could lower solar plant output, such as dust accumulation, vegetation growth or equipment wear, allowing preventive action before productivity drops.
In engineering and materials, L&T said it has built tools that recommend concrete mix compositions based on required strength and site conditions, reducing repeated testing cycles. It is also using computer vision and character recognition to analyse complex piping diagrams and automatically extract quantities of components.
“These are all generic algorithms. Once trained, they can be shared and then used to build (use cases),” Ganesan said.
The broader ambition is to combine such domain-led use cases with investments in data centres, compute capacity and partnerships, creating what the company described as an AI factory ecosystem. “This is one of the unique cases where you can not only take captive use cases to market but also by virtue of having worked with different domains, bring them in,” he said.