A software update can give a vehicle a new capability. The harder question is whether the manufacturer can introduce that change without compromising the systems already keeping the vehicle safe and operational. As commercial vehicles become increasingly software-defined, every new feature brings a corresponding engineering obligation. Now, more than ever, the architecture must accommodate change, cybersecurity must keep pace, and the vehicle must remain reliable throughout its working life.
For Volvo Group, this is where the promise of software-defined vehicles meets the realities of automotive engineering. Artificial intelligence can accelerate code generation and testing, but faster development does not automatically translate into a safer or more dependable vehicle. The greater challenge is understanding how every change affects the system as a whole.
Jaikishan Agnihotri, Head of Connectivity globally and Head of Function for Software Defined Vehicles in India at Volvo Group, sees this distinction shaping the next phase of automotive development. As AI takes on more of the work involved in writing and testing software, he argues, engineers will need to retain a deeper understanding of the vehicle itself. His assessment of AI notes that while AI can accelerate software development and testing, its value ultimately depends on whether it improves the engineering process and produces measurable business outcomes.
The vehicle does not stop evolving at delivery
The shift towards software-defined vehicles changes the relationship between manufacturers and customers. A vehicle is no longer necessarily a fixed product whose capabilities are determined entirely at the time of purchase. Software updates can introduce new functionality and create opportunities to deliver value throughout its operational life.
For Agnihotri, the distinction between the core software platform and its adaptable components is fundamental. The platform must be sufficiently resilient to accommodate repeated updates, while the more flexible elements allow manufacturers to introduce features without rebuilding the underlying architecture.
“The important thing is what is our core in the software platform, and what is adaptable,” he says. “The adaptable part is where we will see more and more updates coming in, the feature updates. Over the years, updates will really play an important role over the life cycle.”
But the proposition goes beyond the technology itself. The commercial model must increasingly account for what happens after a vehicle is sold.
“It’s not about selling vehicles. It’s all about how you maintain that vehicle for a lifetime,” Agnihotri says. “All the software updates are more about how you create the value.”
The qualification is important in commercial transport, where a vehicle’s mechanical life, software lifecycle and operating requirements do not necessarily move in lockstep. Updates must also be secure enough for customers to trust remote delivery.
Agnihotri points out the frequency of updates has to be considered alongside their safety and security. A software platform that can evolve is valuable only if the process of changing it does not introduce unacceptable risks.
Connectivity makes security a continuing engineering responsibility
The more functionality a vehicle derives from connected software, the more consequential its vulnerabilities become. Cybersecurity is not simply about protecting information. In a connected truck, the integrity of the systems governing vehicle functions also has implications for driver safety.
Agnihotri places safety and security at the centre of Volvo Group’s approach, pointing to regulatory requirements as well as engineering methods used to identify potential weaknesses. “Safety and security is something which we can’t compromise as Volvo Group. It is very important for us how we keep the vehicle safe for our customers and how we also keep the drivers safe on our trucks.”
The work includes failure mode and effects analysis, design failure mode and effects analysis, hazard analysis and threat analysis. These methods help engineers assess potential failures and threats before they compromise a system.
The challenge is to maintain that discipline as vehicles become more software-intensive. Frequent updates, expanding connectivity and increasingly complex architectures make security a continuing responsibility rather than a final stage in product development.
AI must demonstrate value beyond code generation
Artificial intelligence is beginning to change parts of Volvo Group’s software development and testing processes. Agnihotri describes applications in development workflows and test automation, including the structuring of test cases. The objective is to improve efficiency, but deployment is not treated as an end in itself.
The company evaluates use cases through proofs of concept and examines whether the technology produces tangible benefits. Development cycle time provides one measure. A process that previously required eight weeks, for example, could become more efficient if the time were reduced to six or four weeks.
Agnihotri says the group has seen savings and positive business outcomes across several applications, although he does not disclose quantified results.
His approach reflects a distinction between adopting AI because it is available and identifying where it can solve a genuine engineering problem. “When AI evolved, in the Indian industry also, it was all about chatbots,” he says. “But what we did differently, I can say, is that we really have waited to see where we can get the business value.”
For software-defined vehicles, that means examining whether AI can contribute across specification, development and verification rather than applying it indiscriminately. It also means evaluating the cost of deployment against the efficiency gained. “You can put AI on everything. But it is important that AI should be used where the business value comes in the best way,” he adds.
The more consequential question, however, concerns the role of engineers as AI takes on a greater share of routine development work.
Why understanding the entire vehicle matters
AI may assist with code generation, but automotive engineering cannot be reduced to producing code. Software operates within a larger system of hardware, interconnected components and functions whose behaviour must remain predictable.
Agnihotri expects the emphasis to shift towards engineers who can understand these interactions rather than work exclusively on individual components. “You can replace coders by putting AI in the software department,” he avers. “But you still need an end-to-end person who knows the system, what is really happening.”
He sees greater importance ahead for vehicle architects and system engineers who understand the entire vehicle and how its constituent parts work together. Volvo Group, he says, has been strengthening its system engineering and architecture capabilities in India.
This is a more nuanced picture than one in which AI simply replaces engineering labour. The transcript does not establish how many roles might be affected, but Agnihotri’s emphasis is clear. As routine coding becomes increasingly automated, engineers will still need to understand the consequences of changes across the vehicle.
That expertise becomes especially important when AI-generated or AI-assisted software must be evaluated against the safety, security and reliability requirements of a commercial vehicle.
From predicting faults to self-healing vehicles
The next application of AI could move beyond development efficiency towards the operation of vehicles themselves. Connected vehicles generate information that could help manufacturers identify emerging problems before they become customer-facing failures.
Agnihotri describes predictive diagnostics as an area of ongoing development, with the longer-term possibility of vehicles being able to correct certain issues themselves. “It is more about how you can monitor issues in the best possible way before they really become a customer issue,” he says.
The ambition extends beyond detecting a fault. “You predict the issue. Can you also help fix that issue on the vehicle? If the vehicle can really self-heal, that is something which you need to see,” he says.
The distinction between prediction and correction remains substantial. Identifying an emerging problem is one task. Determining whether a vehicle can safely resolve it requires confidence in both the diagnosis and the corrective action.
Agnihotri does not offer a definitive deployment date, but says the industry is moving towards more advanced vehicle diagnostics, with developments potentially emerging over the next two to three years.
India’s role must extend beyond execution
The same shift towards deeper system-level expertise has implications for India’s engineering operations. Agnihotri describes Volvo Group’s Indian hub as having evolved beyond an execution-focused role towards greater strategic responsibility across research and development and IT.
Indian teams have responsibilities in software-defined vehicles, connectivity and vehicle security, including contributions to technology roadmaps and product strategy.
The opportunity, he argues, is to expand that ownership into areas where India has the potential to contribute more substantially, including AI, cybersecurity, simulation and virtualisation. “India also has the potential to grow in digitalisation. It has potential to grow in AI,” he says. “India also has a potential to grow more into simulation as well.”
For Agnihotri, the broader direction is captured in a simple formulation. “What we need to build is India for India and India for the world.”
The ambition is not merely to expand the volume of engineering work undertaken in India, but to deepen its responsibility for the technologies that shape future products.
As vehicles become increasingly dependent on software, the engineering challenge will extend beyond developing features or accelerating code generation. It will require an understanding of how software, hardware, diagnostics and security interact throughout a vehicle’s life. AI may help engineers work faster, but the more difficult task remains ensuring that the vehicle as a whole continues to work as intended.