A modern vehicle can accommodate more than 150 ECUs, tens of millions of lines of code, and generate substantial volumes of data. This sophistication fuels innovation, but it also adds to development, validation, and maintenance costs.
For automakers, the economic challenge now is to master this complexity without slowing time-to-market.
Artificial intelligence opens up an intriguing path, because beyond adding new onboard services, it can help engineering teams understand systems that have become hard to grasp.
Connecting Dispersed Information
For a long time, analytics tools focused on structured data: CAN signals, analog data, error codes. Yet, the root causes of a problem are rarely found in those sources alone.
Indeed, identifying the source of a fault also relies on analyzing a technical specification, a test report, a repair history, or a software modification.
All of these elements are often spread across multiple tools, teams, and suppliers. In fact, a large portion of knowledge remains too dispersed—between documents, business tools, and individual expertise. For example, when an engineer leaves a project or a company, much of their knowledge leaves with them.
AI can precisely bring these sources closer together, provide their context, and help engineers formulate more quickly relevant hypotheses. Reducing the time spent searching, comparing, and interpreting information can shorten diagnostic and validation cycles. On the scale of an automotive program, these gains directly impact engineering costs, the utilization and number of prototypes, and the timelines of industrialization.
The Quality of Context Will Determine the Value Created
The performance of an AI system depends less on the raw amount of data available than on its ability to access the right information, at the right moment, with enough context to interpret it.
A single signal ultimately tells very little: its value emerges when it can be linked to the conditions under which it was produced, the vehicle’s state, its history, or other events occurring at the same time.
The challenge for automakers is, therefore, to have ingestion pipelines capable of collecting and dynamically contextualizing the relevant data, rather than trying to centralize all information by default. AI can then cross-reference vehicle data, diagnostics, logs, technical documentation, or maintenance history to reconstruct a much more complete view of an event and steer the analysis toward the most probable causes.
This is precisely where AI changes the scale of analysis. It can process and relate a volume of information far beyond what a human could manually review, and simultaneously surface more correlations, hypotheses, and paths to resolution. The engineer thus gains a richer understanding of the problem and can devote their expertise to interpreting the results and making decisions, rather than the tedious search for the information needed to diagnose.
Ultimately, in automotive competition, the advantage will go to companies able to convert their data and experience into industrial decisions more quickly. AI contributes to that, provided it is viewed as a productivity and knowledge-capture tool, rather than as a technological promise in itself.
*Alexandre Corjon is Senior Vice President of Engineering at Sonatus
