The automotive industry is going through one of its most complex transitions in decades. Established manufacturers are trying to balance the shift to electric and software-defined vehicles (SDVs), against rising software investment, growing competition and the pressure to bring new services to market faster.
The industry’s economics are changing too, as battery costs, tighter margins and lower potential service revenue from electric vehicles are putting pressure on traditional profit pools. While the move towards electric and software-defined vehicles brings commercial opportunity, it also raises the operational stakes. The challenge is no longer just how quickly OEMs can build software into vehicles, but how effectively they can operate, monitor, update and secure them once they are on the road. This is where connected vehicle operations become central to long-term success.
Automotive is moving beyond manual operations
Modern vehicles increasingly rely on cloud platforms, embedded applications, connectivity services, telematics, diagnostics, infotainment systems and over-the-air updates. This creates a more dynamic customer experience, but it also makes vehicles far more complex to operate and support throughout their lifecycle.
When something goes wrong, teams need to quickly understand whether the issue sits in the vehicle, the network, the cloud environment, the application layer or a third-party service. In a large, connected fleet, even a small percentage of vehicles failing to connect can create a significant operational burden.
Software and AI can address these issues, but OEMs have been slow to adopt AI-driven automation to improve operational efficiency. Changing that requires a shift in mindset. Instead of treating AI as a sweeping transformation effort, OEMs need a practical maturity model for applying it to operations. Fortunately, the industry already has a useful framework in place: the staged progression from manual driving to full vehicle autonomy.
A familiar framework for an unfamiliar problem
Levels of driving automation define the journey from Level 0, with no driving automation, to Level 5, where a vehicle can operate fully autonomously under all conditions. That framework works because it shows that full automation is not a single leap, but a staged progression.
Instead of asking whether operations are manual or autonomous, OEMs can assess where they sit on a maturity curve. That matters because operational AI will not mature overnight. It will move through stages, from basic assistance to partial automation, then to conditional autonomy and eventually to highly autonomous operations.
1. Level 0: OEMs are still in manual mode
At Level 0, operational teams are responsible for monitoring, diagnosing and resolving issues manually. This remains common across many areas of automotive IT and operational technology. Teams rely on alerts, tickets, dashboards, logs, and customer reports to understand what is happening across connected-vehicle environments. These manual processes become harder to sustain as vehicle software, connectivity and cloud dependencies increase.
If a group of vehicles stops communicating, the root cause could sit in the vehicle software, SIM profile, mobile network, cloud platform, backend service or application logic. Without unified visibility, teams can spend significant time isolating the problem before they can begin to fix it. This increases costs, slows response times, and reduces the ability to scale connected services efficiently.
2. Levels 1 and 2: AI becomes the operational co-pilot
At Level 1, AI starts to assist operational teams by surfacing patterns, anomalies, and likely causes faster than manual investigation alone can. It might identify unusual traffic behaviour, cluster similar incidents, flag a connectivity issue affecting a specific region or summarise relevant logs across multiple systems. The human team still owns the decision, but AI reduces the time needed to understand the problem.
At Level 2, AI starts to recommend action. It may suggest a likely root cause, prioritise the most urgent issues, recommend the right team to handle the case or propose a remediation step. This is where AI becomes an operational co-pilot. It does not replace human judgement, but it helps teams move from investigation to action more quickly.
For OEMs, the immediate value is practical. Faster diagnosis means lower operational cost, less downtime, fewer escalations and better use of engineering and support resources.
3. Level 3 is where responsibility starts to shift
The most important transition happens at Level 3 – this is where AI moves beyond advice into action and the operating model begins to shift. In driving automation, Level 3 is where the automated system can perform the driving task under certain conditions, while the human remains available to take over. This is a significant change because responsibility begins to move from the human to the system.
The same shift can happen in automotive operations. At Levels 1 and 2, AI assists the operations team. At Level 3, the operations team begins assisting AI. In practice, this means AI can lead defined operational tasks within agreed boundaries. It could monitor a specific service, detect a known class of issues, initiate an approved response or escalate only when confidence is low.
This is not full autonomy. Human teams remain essential for oversight, exceptions, governance and continuous improvement. However, it changes the operating model. Instead of every issue starting with human investigation, AI can begin to manage repeatable tasks while people focus on more complex, higher-value work.
4. High automation changes the economics of SDVs
At Level 4, AI can manage a broader range of operational scenarios with limited human involvement. This could include monitoring connected services, validating OTA update performance, identifyingabnormal device behaviour, detecting connectivity failures or supporting security responses across defined environments.
This level of automation becomes increasingly important as vehicles become more software-defined. As software content grows, operational complexity grows with it. Every new connected feature creates another dependency that must be monitored, updated and secured.
Without greater automation, OEMs risk scaling operational costs alongside software complexity. That is not sustainable in a market already facing margin pressure. High automation allows manufacturers to absorb more complexity without simply adding more manual effort. It also helps improve time to market, because teams can launch, monitor and improve software-enabled services with greater confidence.
5. Fully autonomous operations remain the long-term goal
Level 5 is the long-term vision: autonomous operations that can continuously monitor, diagnose, resolve, and optimise.
In this model, AI agents would not only respond to known issues. They would proactively identify emerging problems, predict failures, coordinate fixes, and optimise performance across the connected-vehicle lifecycle.
Human teams would still play a critical role, but their work would shift towards governance, strategy, exception management, system improvement and accountability. It should mean using AI-driven automation to remove repetitive investigative work, reduce manual intervention, and allow skilled teams to focus on the areas where human judgment matters most.
Visibility comes before autonomy
The truth is that OEMs cannot automate what they can’t see.
Before manufacturers can move towards higher levels of operational AI, they need visibility across vehicles, networks, applications, cloud platforms and security events. Without that foundation, AI systems risk making recommendations based on incomplete or unreliable data. This is particularly important as regulations around vehicle software and cybersecurity become more demanding.
Operational AI also needs strong governance. For OEMs, autonomy must be built on clear escalation paths, explainability, auditability, data quality and security controls.
AI maturity must be staged, not rushed
The automotive industry already understands that autonomy requires stages. No manufacturer expects to move from manual driving to full autonomy in a single step, and AI in automotive operations should be treated the same way.
For many OEMs, the first step will be using AI to help teams diagnose issues faster. The next will be using AI to recommend actions, automate repeatable workflows and manage defined operational tasks.
Over time, as systems become more reliable and governance matures, AI can take on more responsibility. This staged approach turns AI from a broad ambition into an operational roadmap, helping OEMs reduce cost, improve time to market and manage the complexity of SDVs without losing control.
The destination may be autonomous operations, but the value starts much earlier. For OEMs under pressure to innovate faster and operate leaner, even the first steps toward AI maturity can deliver measurable impact. The journey to fully autonomous operations starts today, one level at a time, with the aim of achieving SDV operational efficiency across the lifecycle of the vehicle.
