AI is being democratised at full throttle - much in sync with the government's target of a $1 Tn digital economy by 2030 - and what fuels this growth is the economics of AI that businesses across industries are trying to gain from.

Indian infrastructure major Larsen & Toubro (L&T) made no mistake in reading the future. AI is no longer an experimental capability or a future-facing investment - it is increasingly being embedded into the core mechanics of how the company executes projects at scale.

In an interaction with Inc42, R Ganesan, who heads the Corporate Centre at L&T Construction, framed the company's AI strategy as deliberately multi-layered. "We see different use cases across our various businesses and multiple applications of AI that can impact our execution either in terms of productivity, time to value, autonomous or near-autonomous decision-making and in some cases requiring us to reimagine processes or tasks," he said.

As India emerges as the third most competitive nation in AI adoption, led by its 6 Mn-strong army of technology graduates and powered by an ecosystem of 1.8 Lakh startups and more than 1,800 global capability centres, infra companies are increasingly adopting AI to improve efficiency, safety, and project timelines.

"How you deploy the technology and how you execute it make all the difference in this increasingly competitive landscape," Ganesan said. Instead of treating AI as a single transformation lever, his company is deploying it across the enterprise as both an execution accelerator and a structural capability.

The ambition, according to the executive, is to leapfrog to the next generation of EPC and manufacturing businesses by embedding AI and intelligent automation directly into workflows that define scale, cost, and delivery reliability.

AI In Layers From Tenders To Projects

A defining feature of AI adoption at L&T, Ganesan said, is its breadth across the project lifecycle.

AI systems are already active at the tendering stage, where teams analyse large volumes of contracts, technical documents, and risk variables under tight timelines. The entire process was dependent on manual review and expert judgment, often stretching decision cycles over months until AI shrank the process to days or weeks.

At the back office, AI systems are being used to auto-process several hundred thousand supplier invoices. This was previously manual, time-consuming, and error-prone. Automating this layer has freed up teams to focus on higher-value activities while improving accuracy and throughput.

"Contract risk classification which used to take months to conclude can now be completed within hours," Ganesan said.

The compression of these timelines directly impacts bid responsiveness and reduces friction in early-stage decision-making.

Beyond tendering, AI is being applied across multiple execution stages in both EPC and manufacturing businesses, supporting planning, monitoring, and risk identification in environments where even small inefficiencies can cascade into large cost overruns.

At the project level, AI is being used to press warning signals, point bottlenecks and predict risks. It helps teams preempt a challenge and prepare for it. In other words, being proactive, instead of being reactive after the process is paused.

Safety remains another critical focus area for the deployment of AI. Across multiple project sites, AI-powered computer vision systems analyse live camera feeds to flag safety violations almost in real time. The company is now extending these capabilities by integrating AI with robotics and drone technologies to explore safety capabilities.

While construction has historically faced steeper adoption challenges for physical AI because of its complexities and unit economics, L&T addresses this through what Ganesan described as a holistic approach aimed at adoption at scale and a hybrid build-and-buy approach.

By integrating AI applications, machinery and workforce planning processes have been reduced to 10 minutes from 2 weeks earlier. Predictive analytics and process optimisation helped L&T reduce project costs by 2-3%, contributing to enhanced operational productivity.

When compared to players like TATA Projects, unlike L&T's focus on a custom AI layer (LNTCS), they have anchored their transformation on SAP Business Suite to unify key business functions on a single, intelligent platform.

"Our LNTCS platform hosts and provides role-based access to different AI-based solutions that cut across the lifecycle of various businesses," Ganesan said.

The Platform And A Push Out Of Legacy

Under the hood, L&T operates a deliberately hybrid AI infrastructure.

The group runs AI workloads across on-premise and cloud environments, selecting deployment models based on data sensitivity, latency, and scale requirements. "L&T exercises both its on-prem and cloud environments for running various AI workloads," Ganesan said.

Model-wise, the company uses a wide spectrum - from traditional ML libraries to open-source models such as Llama, besides proprietary offerings like Azure OpenAI.

L&T centralises its AI ecosystem through LNTCS, the company's proprietary cognitive services platform. By hosting all solutions here, the platform functions as a vital control plane that provides a unified experience layer, standardised authentication, and streamlined service management. It allows an organisation to maintain strict governance while ensuring consistent deployment across business sectors.