The AI boom has largely been defined by chatbots and large language models. But in the energy sector, where operators manage complex industrial assets worth billions of dollars, companies are pursuing a different vision of AI—one built around engineering, physics, and operational reliability rather than conversational interfaces.
That vision is driving INSITE 3.0, a platform developed by KBR and Applied Computing. KBR, the US-headquartered engineering and technology company, works across energy, chemicals, defence and infrastructure, while Applied Computing is a London-based AI company founded by former Shell executive Dan Jeavons.
Together, they have created a platform that combines operational data, engineering simulations, and AI models to help industrial operators optimise performance, improve energy efficiency, and reduce emissions.
The collaboration also highlights a broader trend: Bengaluru is now becoming an increasingly important for building industrial AI systems.
Why Energy Needs a Different Kind of AI
Unlike consumer applications, industrial environments leave little room for error. A refinery operator cannot rely on an AI system that is only partially accurate when decisions affect safety, production, and asset performance.
"The biggest challenge that you see is bringing together two worlds that are very different," says Dan Jeavons, President of Applied Computing, in a conversation with AIM. "On the one hand, the world of AI… is moving very quickly, and then [we’re] trying to integrate that with the world of energy, which is very specific." He argues that while generic AI models can tolerate [some] inaccuracy, industrial applications demand far greater precision.
That requirement has led KBR and Applied Computing to focus on what they describe as physics-informed AI. Rather than relying solely on large language models, the platform incorporates engineering simulations and operational knowledge to understand how industrial systems behave in the real world.
"We will not use large language to do engineering decision-making. It's not the right tool," notes Andy Webster, Senior Director of Digital of Sustainable Technology Solutions at KBR. "We hurt people if we get things wrong." The aim, he says, is not to replace engineers but to provide tools that support better decisions in safety-critical environments.
For KBR, the initiative is more than a technology experiment. Roderick Macdonald, Vice President of KBR, India, describes it as central to the company's long-term strategy. "We are a traditional energy-based engineering firm, so the level of investment and trust that we have in this is actually fundamental to where we see the future of the company," he says.
Data, Trust, and the Challenge of Scale
Building AI for industrial operations presents a different set of challenges from those faced by consumer technology companies. One of the biggest obstacles is the data to train AI models.
Energy facilities generate enormous amounts of information, but much of it reflects stable operating conditions because plants are designed to run continuously. As Webster notes, operators often provide the same types of data repeatedly, making it difficult for AI systems to learn how assets might behave under challenging circumstances. To overcome this, KBR combines operational data with engineering simulations that help create a broader picture of possible operating scenarios.
Accessing data is another challenge. Jeavons notes that there is very little public information available because operators regard infrastructure data as highly sensitive. Legal considerations, therefore, became a major part of the project, with KBR implementing strict controls around how customer information is used and shared.
Applied Computing has built dedicated cybersecurity capabilities to protect customer environments, recognising that operators will only adopt AI if they are confident their data remains secure.
In sectors where facilities can operate for decades and experienced operators have spent entire careers managing them, earning that trust may be as important as the technology itself.
Improving Efficiency and Supporting Energy Transition
The immediate objective of INSITE 3.0 is to help operators run facilities more efficiently.
According to Jeavons, reducing energy intensity is one of the most common customer requests. Even a 1–3% improvement can produce significant savings because refineries, petrochemical plants, and similar facilities rank among the most energy-intensive operations in the world, consuming more than 10% of the world’s energy end-use, per the International Energy Agency. Lowering the energy consumption will not only reduce costs but can also substantially reduce emissions.
Macdonald also points to opportunities in areas such as carbon capture, where AI could help improve the performance of emerging technologies.
Beyond operational improvements, KBR sees potential for AI to transform how industrial projects are designed and delivered. Macdonald describes that as the "real prize" for the business. "This is going to transform the way that we engineer and design plants," he says, pointing to a future where engineering workflows can be accelerated and new technologies brought to market more quickly.
Why Bengaluru is Becoming the Nerve Centre
Although KBR and Applied Computing are headquartered elsewhere, both companies see Bengaluru as a critical part of their industrial AI strategy.
Jeavons points to the city's unique combination of engineering expertise, software talent, and research institutions. He highlights the strength of the local talent pool and institutions such as the Indian Institute of Science, arguing that few locations offer the same concentration of skills needed to develop industrial AI systems.
The Karnataka capital’s advantage extends beyond talent. The executives repeatedly emphasise speed. Jeavons notes that the companies established a new innovation facility in less than eight weeks, something he believes would be difficult to replicate in many Western markets. In a field evolving as rapidly as AI, that ability to move quickly is becoming a competitive advantage.
Macdonald argues that Bengaluru is part of a broader Indian engineering ecosystem that will play an important role in industrial AI. Combined with growing energy markets across India, Southeast Asia, and the Middle East, the city offers companies a strategic base from which to develop and deploy new technologies.
India's industrial sector could be among the biggest beneficiaries of platforms such as INSITE 3.0. The country is expanding refining and petrochemical capacity, projected to grow at 5.2% and 7.22%, respectively, per Mordor Intelligence. It is also investing in low-carbon technologies and pursuing ambitious industrial growth plans. These are precisely the kinds of industries where AI-driven optimisation can deliver measurable gains.
Your reaction