By Raj Kanaya, General Manager, Automotive Business Unit, Aeris
The automotive industry is rapidly evolving into a more connected, software-defined future. This shift is creating a new challenge for OEMs as vehicles become increasingly software-led, Cloud-connected and reliant on real-time diagnostics, OTA updates and always-on connectivity.
This opens new opportunities to improve driver experiences and accelerate software innovation, but it also creates a more complex environment to manage. A vehicle is no longer a static product that changes only during scheduled service visits; it is part of a wider IoT ecosystem, where data flows continuously between the vehicle, network, Cloud platforms, applications and service providers.
For OEMs, the challenge is no longer simply how quickly they can build software into vehicles. It is how effectively they can operate, monitor, update and secure connected vehicles once they are on the road.
Connected vehicles are creating a visibility problem
Connected vehicle data has huge potential, but only if OEMs can interpret and act on it quickly.
When an issue occurs, teams need to understand whether the root cause sits in the vehicle, an IoT device or sensor, SIM profile, mobile network, cloud environment, application layer or a third-party service. At scale, even a small percentage of vehicles failing to connect can create a significant burden, with teams spending days identifying the issue before they can fix it.
The problem is not necessarily a lack of data, but that it is fragmented across different systems, partners and operational teams. This is where AI becomes operationally valuable. It can help OEMs move from reactive troubleshooting to more proactive, data-led decision-making. Rather than relying solely on manual alerts, AI can analyse patterns across multiple sources and help teams understand what is happening more quickly.
Why OEMs need an AI maturity model
The automotive industry already has a useful framework for understanding vehicle autonomy: the six levels of driving automation, from Level 0 to Level 5. The same principle can be applied to connected vehicle and IoT operations.
● At Level 0, OEMs rely on manual investigation, with teams piecing together alerts, logs and dashboards to determine what has gone wrong.
● At Levels 1 and 2, AI starts to assist by identifying patterns, prioritising incidents and recommending likely causes across vehicle, IoT device, network and application data.
● By Level 3, responsibility begins to shift, with AI taking the lead on specific tasks while humans supervise and intervene as needed.
● Levels 4 and 5 represent the longer-term goal: highly automated, and eventually autonomous, operations that can monitor, diagnose and optimise with minimal manual intervention.
Progress will come from building confidence in specific use cases, such as diagnostics, OTA assurance, anomaly detection and connected service monitoring.
AI can accelerate diagnostics
Modern vehicles produce a significant amount of operational data across onboard sensors, IoT devices, vehicle performance, connectivity and software behaviour. The value lies in connecting those signals quickly, so teams can identify likely causes faster.
For example, if vehicles in one region begin showing connectivity failures, or several vehicles display similar errors after a software release, AI could help teams determine whether the issue is linked to a network outage, device configuration, backend service, software fault or application behaviour.
This does not remove the need for human expertise, but it can reduce the time spent identifying the source of a problem, lowering operational costs and helping teams move faster from investigation to response.
OTA updates need operational intelligence
OTA software deployment is becoming critical to the software-defined vehicle lifecycle. OEMs need to deliver patches, new features, performance improvements and security updates remotely across large numbers of vehicles.
However, OTA updates depend on reliable connectivity, IoT device readiness and clear feedback from the vehicle estate. If an update fails, teams need to know why. AI can help by monitoring OTA performance, identifying failed or incomplete updates, spotting repeated transmission attempts and highlighting where connectivity issues may be affecting deployment.
Failed updates are not just a technical inconvenience. They can increase cost, delay feature rollouts, create customer experience issues and leave vulnerabilities unresolved. Understanding update performance in near real time also helps OEMs identify which vehicles, markets, networks or configurations are most likely to experience issues, and adapt deployment strategies accordingly.
Cybersecurity is becoming part of connected vehicle operations
As vehicles become more connected, their potential attack surface grows across networks, applications, APIs, software components and cloud services. Cybersecurity must therefore be integrated into operations, with OEMs able to detect abnormal behaviour early and act before issues affect a wider fleet.
AI can support this by analysing traffic patterns, IoT device behaviour, and operational signals across the connected-vehicle ecosystem. It can help flag anomalies that may indicate compromised endpoints, malware, misconfigurations or unexpected data usage.
For example, a sudden increase in data usage or repeated failed connection attempts could signal an operational fault or a potential security risk. AI can help teams prioritise these signals and escalate the right issues faster. This is especially important as automotive software, connectivity and cybersecurity requirements become more demanding. OEMs will need stronger visibility, auditability and response capabilities across the full vehicle lifecycle.
From assisted intelligence to agentic AI
As OEMs build confidence in specific use cases, operational AI can move from identifying issues to recommending, coordinating or initiating action within defined boundaries.
This is where agentic AI becomes relevant: systems that can work towards a specific operational goal, such as diagnosing a connectivity issue, monitoring service performance or coordinating a known response, while still operating within clear rules and human oversight.
As these systems mature, they could suggest remediation steps, automate repeatable workflows and escalate issues only when human intervention is needed. This should not mean removing people from the process. Human teams will remain essential for oversight, governance, exception handling and continuous improvement. The real opportunity is to reduce repetitive investigative work so that skilled teams can focus on higher-value decisions.
Operational AI depends on connected vehicle data
AI adoption in automotive cannot be separated from the broader connected vehicle and IoT ecosystem. AI is only useful if it has access to reliable, timely and contextual data from across vehicles, networks, applications and cloud services.
The manufacturers that make the most progress will be those that treat AI not as a standalone technology but as an operational layer across the connected-vehicle lifecycle. That means using AI to improve diagnostics, accelerate OTA deployment, strengthen cybersecurity and optimise performance at scale.
The destination may be more autonomous operations, but the value starts much earlier. By applying an AI maturity model based on the same staged logic used for driving automation, OEMs can move from manual investigation to assisted intelligence, then towards more autonomous operational models. Even the first steps towards AI-led operational intelligence can help OEMs reduce costs, improve resilience and bring new services to market faster.
Author biography:
Raj Kanaya is General Manager, Automotive Business Unit, Aeris
