As AI alters vehicle testing, engineering context will determine its impact. Fernando Valera, CTO at Visure Solutions, discusses the company’s AI agents and how they provide governed engineering context that improves agility, reduces rework, strengthens governance and accelerates product delivery
Artificial intelligence is becoming an integral part of automotive development, promising faster test generation, automated analysis and more efficient validation workflows. Yet as vehicles become increasingly software-defined, the real challenge is no longer adopting AI, it is ensuring AI can be trusted within safety-critical engineering processes. The next evolution of automotive testing will depend on combining AI with engineering context, governance and end-to-end traceability.
The automotive industry is undergoing a profound transformation. Software-defined vehicles, ADAS, electrification, connected services and over-the-air updates have dramatically increased software complexity while shortening development cycles. Every software release must be validated against thousands of requirements across multiple vehicle domains and stringent functional safety and cybersecurity standards. As a result, verification and validation (V&V) teams are under growing pressure to accelerate testing without compromising quality or compliance.
Artificial intelligence is helping address this challenge. Engineers are already using large language models to review specifications, generate test cases, summarize technical documentation and support engineering analysis. While these capabilities improve productivity, they expose a critical limitation: generic AI understands language but not the engineering relationships that determine whether a vehicle is safe, compliant and ready for release.
AI needs engineering context
Automotive testing is built on traceability. A seemingly small software modification can affect requirements, system architecture, safety goals, cybersecurity controls, verification procedures and regression test suites. Determining those impacts requires understanding how engineering artifacts are connected throughout the product lifecycle.
Generic AI has no inherent awareness of these relationships. It may recommend a new test case or rewrite a requirement, but it cannot determine whether verification remains complete, whether a safety requirement is still satisfied, or whether a change introduces gaps in compliance. Those answers exist within the engineering lifecycle, not inside the AI model.
This challenge is becoming even more significant as SDVs receive continuous software updates throughout their lifetime. Verification teams need AI that can reason using trusted engineering information rather than isolated documents or manually supplied prompts.
Engineering intelligence for modern verification
This need is driving the emergence of engineering intelligence, an approach that combines AI with engineering knowledge, lifecycle relationships and traceability.
Instead of treating AI as a standalone assistant, engineering intelligence connects it to requirements, risks, test cases, defects, software changes and verification evidence. AI can then analyze engineering information in context, helping teams identify affected test cases, locate missing traceability, evaluate change impacts, and detect potential coverage gaps before they become costly defects.
One area where this delivers immediate value is regression testing. As software updates become more frequent, rerunning every test after every change is increasingly impractical. AI that understands engineering relationships can prioritize regression testing based on actual system impact, reducing unnecessary validation effort while maintaining confidence in software quality and functional safety.
Engineering intelligence also improves collaboration across systems engineering, software development, testing, cybersecurity and functional safety. Instead of searching across disconnected tools, teams gain faster access to the engineering knowledge needed to make informed decisions.
Trust requires governance
For safety-critical automotive systems, trustworthy AI depends as much on governance as it does on intelligence.
Standards such as ISO 26262, ASPICE, ISO/SAE 21434, and ISO 21448 (SOTIF) require engineering decisions to remain traceable, reviewable and supported by objective evidence. AI-generated recommendations must therefore remain linked to authoritative engineering data and subject to human review.
Organizations also need clear policies defining what information AI can access, how engineering data is protected, and how AI-generated outputs become part of existing approval workflows. Human oversight remains essential, not because AI lacks value, but because engineering accountability cannot be delegated.
Connecting AI to the engineering lifecycle
Providing AI with meaningful engineering context requires secure access to lifecycle information rather than isolated documents. Technologies such as the model context protocol (MCP) provide a standardized way for AI systems to retrieve governed information directly from engineering applications.
Instead of manually copying requirements, test specifications or safety analyses into external AI tools, engineers can allow AI to securely access approved engineering information while respecting permissions and organizational governance. This enables AI to analyze requirements alongside verification evidence, understand relationships between tests and risks, and support impact analysis without disrupting established engineering workflows.
Engineering intelligence in practice
Engineering Intelligence is already moving from concept to implementation. Visure Solutions supports this approach through Vivia, its AI assistant, and the Visure MCP Server.
Vivia provides contextual AI assistance directly within the engineering lifecycle, helping teams improve requirements quality, analyze traceability, assess change impacts and support verification activities using the engineering information already available within the project.
The Visure MCP Server extends these capabilities by securely connecting enterprise AI platforms to governed engineering information managed within Visure. Rather than relying on isolated prompts, AI retrieves trusted lifecycle data directly from engineering repositories while respecting organizational permissions, approvals and audit requirements. This enables organizations to adopt their preferred AI technologies while ensuring responses remain grounded in authoritative engineering knowledge.
Looking ahead
The future of automotive testing will not be defined simply by faster automation or more powerful AI models. It will be defined by how effectively AI can support engineering decisions within the rigorous processes that ensure vehicle safety, quality and compliance.
Engineering intelligence provides that foundation by combining AI with engineering context, traceability, governance and human oversight. For automotive verification teams, this means faster impact analysis, smarter regression testing, higher-quality requirements and more efficient validation, all while preserving the discipline required to develop safe, reliable software-defined vehicles.
Watch this webinar on engineering intelligence: High-Velocity Engineering with Visure MCP: AI Agility, Traceability & Governance
