THE CAR AS AN AGENT – Intent, Experience and the New Automotive Economy

THE CAR AS AN AGENT

Intent, Experience and the New Automotive Economy

Keynote address | India

Distinguished leaders of India’s automotive industry, CEOs, technology leaders, engineers, entrepreneurs, innovators, ladies and gentlemen,

Something fundamental is happening to the automobile.

For much of the last decade, we spoke about the transition from the internal-combustion engine to electric mobility. That transition is still underway. Alongside it, however, another transformation is taking place and it may ultimately be even more consequential.

The automobile is becoming an AI-defined machine.

Look at where some of the world’s most ambitious companies are investing. NVIDIA is building the computing architecture for increasingly intelligent and autonomous vehicles, connecting accelerated computing, simulation and AI from the data centre to the car. Google is bringing Gemini into the automobile, moving interaction beyond rigid voice commands toward more natural conversation, contextual understanding and intelligent assistance.

Mercedes-Benz is developing MB.OS, a software-defined, chip-to-cloud architecture connecting vehicle functions and digital services, while working with leading AI and computing partners. BYD is pushing intelligence across the automotive stack, from advanced driver-assistance and vehicle architectures to AI-enabled development and manufacturing. Tesla has taken another path, treating its vehicle fleet as part of a large learning system and developing end-to-end AI for driving alongside its own inference compute.

These are different companies, pursuing different architectures and different business models. Beneath those differences, however, we can see the same strategic movement: compute is moving into the vehicle, software is becoming continuously upgradeable, AI is becoming multimodal, driving systems are becoming more intelligent and the cabin more conversational.

This is already transforming automotive engineering. More importantly, it raises a bigger question: what happens when the car begins to understand not only the road, but the person, their context, preferences, purpose and, eventually, within boundaries established by us, their intent?

That is where I believe the next automotive transformation begins.

And that leads me to a thought that may sound provocative: we may be approaching the end of the dashboard. Not the end of the screen or of beautiful automotive design, but the end of the idea that humans must continuously tell machines how to serve them.

For more than a century, we designed automobiles around controls: first buttons, dials and switches; then screens, menus and applications. The technology changed enormously, but one assumption remained largely intact: the human operates the machine. Artificial intelligence is beginning to change that relationship. Agentic AI could change it profoundly, because a machine that understands our intent – not merely our commands – can increasingly help determine the steps required to achieve an outcome. When that happens, we do not simply redesign the dashboard. We redesign the relationship between human and machine, the customer experience and potentially the economics of the automotive industry itself.

FROM INTERFACE TO INTENT

Imagine tomorrow morning. You enter your car and say: “I’m running late. I need to be at the airport by six. Pick up my colleague on the way, make sure we have enough charge when we arrive, and please give me twenty quiet minutes to prepare for my meeting.”

Today, delivering that experience might require navigation, calendar, traffic and charging applications, messages, payments and a series of interruptions. An increasingly intelligent vehicle could bring those elements together. It could understand your destination and schedule, account for traffic and battery levels, coordinate charging, and replan when circumstances change. Within the permissions you establish, it could take appropriate action. You have not programmed every step of the journey; you have expressed an intention.

That is the first transition: interface to intent. For decades, we have made technology easier for people to operate. We may now be entering an era in which technology increasingly learns how to operate around people. And that changes what personalization means.

WHEN PERSONALIZATION BECOMES UNDERSTANDING

Today, a personalized car remembers my seat position, temperature, music and driving mode. Useful, certainly, but that is largely configuration. Agentic AI offers the possibility of moving from configuration to preference, from preference to context, from context to intent, and eventually, within clear boundaries, to agency.

Consider two journeys in the same vehicle. On Monday morning, I am late, my first meeting has moved, traffic is unusually heavy and I have an important presentation in forty minutes. The useful response is not another stream of notifications or ten questions. It is assistance that handles what it can and protects the time and attention I need.

On Saturday, the same vehicle carries my family. The destination, entertainment, comfort and charging priorities may be completely different. For an elderly passenger or a person with a visual, hearing or mobility impairment, the interface may need to adapt further, reducing complexity and increasing independence. Personalization then becomes more than a luxury feature; it becomes a means of inclusion.

The aspiration is not technology constantly reminding us how intelligent it is. It is technology that understands what this person needs, in this moment and knows when to remain silent.

FROM ONE MILLION VEHICLES TO ONE MILLION RELATIONSHIPS

Mass production gave the automotive industry scale, consistency, quality and affordability. Artificial intelligence may enable a different kind of scale: individualized experiences for millions of customers. One driver values performance, another accessibility, another family convenience or energy efficiency. Even the same person has different priorities on a working morning, a family holiday or during an emergency. The customer is not a static profile; the customer has a changing context.

Now imagine two manufacturers, each selling one million excellent vehicles. One continues to manage those vehicles primarily as products. The other also builds one million evolving relationships. Software improves after purchase. Maintenance becomes more predictive. Charging and services become more contextual. Partners contribute capabilities that were not present when the vehicle left the factory. The number of cars is identical, but the opportunities to create value across their useful lives may be very different.

This does not mean finding more ways to charge customers for every kilometre. It means creating more value in every kilometre. Customers will reward experiences that make their lives simpler, safer, more accessible and more productive, not intrusive subscriptions or unnecessary interruptions. Better understanding can enable better experiences; better experiences can earn trust; and trust can support deeper, longer-lasting relationships. The reverse is equally true: poor recommendations, unwanted personalization or misuse of data can destroy that trust.

FROM PRODUCT TO PLATFORM, AND FROM VALUE CHAIN TO VALUE NETWORK

The personal computer became a platform. The smartphone became a platform. Cloud computing enabled entire ecosystems. The automobile is different: it is a safety-critical physical system. Yet an intelligent service architecture can increasingly develop around it, connecting software, compute, AI, identity, payments, applications, developers and partners.

That presents automotive leaders with strategic choices. What must we own? Where must we differentiate? What should remain open? Where should we partner, and where should others be free to innovate around us? No manufacturer needs to build everything. In some areas, trying to do so may become a disadvantage.

The industry has mastered one of the world’s most sophisticated value chains. The next automotive economy may also require a value network: automakers working with semiconductor providers, AI companies, energy and charging networks, insurers, banks, mobility services, cities, startups and developers. Each can contribute to the customer experience. The strategic question is no longer only who builds the vehicle, but who creates value around the customer and who participates in that value.

This shift is already visible in industry strategy. PwC finds that almost two-thirds of automotive companies are pursuing ecosystem participation, rising to four in five among the group it identifies as the most future-fit. Those companies are also more likely to have mature software and AI capabilities and to be expanding beyond the traditional automotive offer. The competitive perimeter is moving beyond the vehicle itself.

The risk for automotive is not simply that technology companies might build better cars. It is that another company could become the intelligence layer through which customers experience their cars. The question is whether manufacturers will help shape that layer or leave the relationship to others.

THE BUILDING BLOCKS ARE ARRIVING

At this point, we should distinguish clearly between what is already happening and what may become possible as these capabilities converge. Agentic AI is not simply a concept being discussed for the next generation of vehicles. Intelligence is already moving through the automotive value chain – from research and development and procurement to manufacturing, supply-chain management and the vehicle itself.

The numbers matter. PwC finds that 47 percent of automotive companies are already using AI and advanced technologies across the value chain, a figure it expects to reach 72 percent by 2030. KPMG reports that 86 percent of OEMs are investing heavily in AI, with research and development and supply-chain optimisation among the principal expected gains. ABI Research forecasts shipments of vehicles equipped with agentic AI rising from around five million in 2025 to almost seventy million by 2035. Gartner, meanwhile, identifies agentic AI and physical AI among the technologies reshaping supply chains.

So this is not science fiction. Nor does it mean that every capability I have described is commercially mature. It means that the direction of travel is becoming visible and that the industry is moving from experimentation toward deployment.

That brings us to the larger strategic question. The opportunity is not simply to put another intelligent assistant inside the vehicle or another AI tool inside the factory. It is to determine who will orchestrate the experience around the vehicle and who will create value across the ecosystem that experience connects.

How will these technologies work together? Which architectures, partnerships, standards and trust models will allow them to do so safely? The future is not predetermined. What follows is not a claim that a mature agent-to-agent economy already exists; it is a plausible direction that deserves serious preparation now.

PERSONAL AGENT MEETS VEHICLE AGENT

Imagine that, several years from now, you have a personal AI agent that understands your calendar, preferences, travel patterns and the payment permissions you choose to grant it. You enter a vehicle with its own agent, one that understands the battery, sensors, safety systems, maintenance requirements, traffic, infrastructure and mobility services. Your personal agent understands what you want to achieve; your vehicle’s agent understands the journey and the physical environment.

Now the two systems begin to cooperate. Which recommends a charging provider? Which selects entertainment, communicates with an insurer, initiates an authorized payment or shares information? Above all, which represents your interests?

For years, automotive manufacturers and technology companies have competed over the dashboard, operating system and digital interface. The next competition may concern something more consequential: who earns the trust to orchestrate human intent. That does not necessarily mean owning every service or controlling the customer. It means earning a place in an environment where intelligent systems can collaborate securely and in the customer’s interest. That opportunity is bigger than infotainment and potentially bigger than the car itself.

FROM AGENT-ASSISTED TO AGENT-TO-AGENT COMMERCE

Consider a charging stop. Today, I search, compare, select, navigate, authenticate and pay. An intelligent vehicle could increasingly help with those steps. It knows the battery state and route, identifies compatible infrastructure and recommends options. With additional authorization, it might reserve a charger or execute a permitted transaction. The same logic could eventually extend to maintenance, parking, energy, insurance, mobility services and hospitality.

We should not assume that every one of these interactions will develop in the same way, or at the same pace. Identity, cybersecurity, interoperability, liability, consumer protection and payments all present substantial challenges. But as these systems mature, agent-assisted commerce may develop into more autonomous agent-to-agent interactions. The distinction matters: this is not commerce without people or machines operating without boundaries. Humans establish objectives, permissions and limits; intelligent systems help orchestrate the complexity.

If that model develops, the automobile could become more than a place where commerce is displayed. It could become a trusted participant in how services are discovered and coordinated around mobility. That creates opportunity but it makes trust fundamental.

TRUST BECOMES ARCHITECTURE

A vehicle that recommends a charging station is one thing. A vehicle authorized to negotiate a transaction, share personal information or make a payment carries a different level of responsibility. What can it remember, share, spend or decide? Can its decisions be reviewed or reversed? Who is accountable when something goes wrong?

These are not questions to be added at the end by legal and compliance teams. They must shape the architecture itself: identity, cybersecurity, privacy, authentication, transparency, meaningful permissions, human oversight and accountability. An intelligent vehicle must understand not only what it can do, but what it is authorized to do.

Trust is therefore more than a compliance requirement. It becomes part of the product, the brand and potentially the business model. Trust, however, also requires restraint. Much of today’s digital economy competes for our attention with notifications, messages and recommendations. Perhaps the next generation of premium experience will do the opposite. It will know when to recommend, when to act, when to ask and when to remain silent. The ultimate luxury in the age of AI may be technology that knows when to disappear.

HUMAN+

There is an important principle beneath all of this. Greater machine agency should not mean less human authority. It should mean greater human capability. If an intelligent vehicle can remove unnecessary complexity, return time to a professional, increase independence for an elderly person or someone with disabilities, or make mobility simpler for a family, then AI is doing something more valuable than automating a task. It is expanding what people are able to do.

That is what I mean by Human+. The machine can increasingly manage complexity; the human must retain authority over purpose, permissions and consequential decisions. The more capable these systems become, the more important clarity, privacy, security and accountability become.

AI can orchestrate the journey. Humans retain authority over the destination.

WHY INDIA MATTERS

India brings together automotive engineering, manufacturing capability, software expertise, AI talent, digital infrastructure, entrepreneurial energy and an expanding electric-mobility ecosystem. Just as importantly, it offers something especially valuable in the AI era: complexity at scale. Dense cities, highways and rural communities; different languages, income levels, accessibility needs and infrastructure conditions: solutions developed here must confront complexity as well as scale.

That diversity is an opportunity. If the future is personalization at scale, India offers an important environment in which to develop and test it. Solutions cannot work only in ideal conditions. They must be affordable, reliable and useful to different people in the real world. That creates an opportunity to develop not only vehicles, but also the intelligence, experiences, services, platforms and business models surrounding them.

This will require collaboration between manufacturers, technology companies, startups, researchers, infrastructure providers and government. The ambition can be bold: design it in India, engineer it in India, build its intelligence in India, personalize it at Indian scale, validate it against Indian complexity—and scale it from India to the world.

THE NEXT AUTOMOTIVE ECONOMY

Ladies and gentlemen, for more than a century the automotive industry has asked how to build a better machine. Then it asked how to build a better digital experience. Agentic AI introduces another question: how do we build an automobile that understands the person, their context, intent, changing needs and, crucially, the boundaries they establish?

That is a transition from vehicle to experience, from transaction to relationship, from customer data to customer understanding, and from value chain to value network. The industry has mastered the machine and is mastering software. Its next challenge may be earning the trust to connect people, machines and mobility.

Some elements of this future are already emerging; some are being engineered today; others remain ahead of us. Their convergence is the strategic issue. The time to think about that convergence is not after business models have changed, but before, not because we know exactly what the automobile will become, but because we have an opportunity to shape what we want it to become.

If we get that right, the car can become an agent; the agent can create an experience; the experience can create a relationship; the relationship can enable a platform; and the platform can support an ecosystem and, ultimately, a new economy.

For more than a century, the automobile has given humanity freedom of movement. Perhaps its next chapter can offer something equally precious: freedom from unnecessary complexity. The most sophisticated technology should not constantly remind us that it is there. It should understand us, assist us, protect us and sometimes simply get out of the way.

The most sophisticated technology should not demand more of our attention. It should give attention back to us.

India has the engineering, scale, software, talent, diversity and entrepreneurial ambition to help define this next automotive economy.

Not simply for India. From India, for the world.

Thank you.