When engineers and AI agents work together, engineering is reinvented.
A phase of profound transformation is underway in the world of engineering. Indeed, with the emergence of AI agents, machines go beyond their status as simple tools to become real thinking partners.
In sectors as complex and regulated as mechanical engineering, the automobile or aeronautics industry, this development marks a paradigm shift. Compartmentalized processes are now being replaced by increasingly interconnected and modular systems that automate repetitive tasks and provide valuable support to development teams.
From automation to support
AI and engineering? Nothing new. However, agentic, autonomous and contextual AI is significantly changing expectations. It is no longer a question of automating certain work steps, but of understanding and supporting processes as a whole. These AI agents learn from data, interpret correlations, formulate propositions, and act based on defined goals. They do not replace engineers, but assist them where traditional approaches reach limits in terms of analyzing complex system architectures, evaluating variants, complying with regulatory requirements or managing multiple configurations.
Far from automation in the classic sense of the term, we are talking about a new form of digital assistance, integrated, traceable and scalable.
Why are new solutions needed?
The demands on engineering teams are intensifying. Multiplication of product variants, shortening innovation cycles, strengthening regulatory constraints, growing expectations in terms of quality and customization… All these factors are putting traditional work processes under strain. In the context of software-defined vehicles (SDV), this development is striking, because current development relies on data, interfaces and the ability to update in real time.
This is precisely where the weakness lies. Many development processes still rely on document-based approvals, manual review procedures and insufficiently interconnected tools. Result ? Breaks in continuity, version conflicts, extended feedback loops and, ultimately, innovation dynamics slowed down or exposed to errors.
A response to engineering tensions
AI agents offer a structured response to these challenges and take on new roles in modern PLM, ALM and CAD environments. By analyzing requirements, test reports or configuration states, they identify patterns indicative of risks, opportunities or adjustment needs. On this basis, they make recommendations, prioritize and highlight possible conflicts of interest.
In model-based engineering environments such as Model-Based Systems Engineering (MBSE), AI systems evaluate different design options and simulate their impacts on performance, compliance, and cost before any physical prototyping. This is how to shorten iteration cycles and identify risks early.
In dynamic development projects, agents analyze the consequences of planned developments on existing requirements, components or tests. They propose implementation pathways and assess regulatory risks, thus facilitating informed decision-making. In the case of critical systems where security is a non-negotiable issue (think of vehicles or planes), complete traceability is imperative. AI agents verify the feasibility of new configurations, their compliance with standards, and the completeness of the documentation intended to prove this compliance. Saving time and saving resources.
The intelligent product life cycle as a foundation
To operate effectively, AI agents require a solid database. This is the role of the Intelligent Product Lifecycle (IPL), a structured and semantically coherent data architecture that connects relevant information throughout the lifecycle of a product. From initial specifications to decommissioning, from software versions to maintenance history, IPL creates the necessary conditions for AI to work not only locally, but also systemically.
IPL enables continuous traceability and automated version management. It establishes transversal transparency between teams and tools and guarantees contextual assistance thanks to AI agents. It also facilitates rapid and duly documented decisions in accordance with regulatory processes. Crucial advantage in sectors with high safety requirements.
Concrete examples from industry
Agentic AI already has very concrete applications in industry. Let’s look at some use cases illustrating its value.
● Development of control units: in the automotive industry in particular, AI agents analyze the consistency between software versions, requirements and test cases. BMW, for example, uses AI to automatically check the plausibility and eligibility for approval of combinations of variants.
● In the manufacture of aerospace cables: AI systems detect irregularities in color, insulation or diameter. They compare this data in real time with the digital twin of the product. Errors are thus detected and corrected early.
● In mechanical engineering: the field of plastic injection molding offers us a beautiful illustration. AI-powered validation tools reduce the number of iterations required before mass production by up to 40%. The result is a significant improvement in quality, deadlines and resource utilization.
When humans orchestrate this reinvented engineering
All this transforms the distribution of roles within engineering teams. Design engineers, systems architects or test managers are moving towards digital process orchestration functions. They define rules and control mechanisms, maintain semantic models, configure and train agents, then interpret their recommendations on a global scale.
The framework of traditional technical expertise is then outdated. It requires a systemic approach, an understanding of the domain, a mastery of data and tools, as well as an ability to organize collaboration between humans and AI systems. It is under these conditions that the potential of agentic approaches will be fully exploited.
Focus on trust and traceability
Transparency is key to integrating agentic AI into engineering. Decisions made by AI agents must be traceable and verifiable, both for developers and for audit and governance services. Unquestionably, transparency of IPL is an essential prerequisite.
To be deployed on a large scale, in contexts where security, quality and compliance are critical, agentic AI cannot be perceived as a “black box”, but as an explainable and documented assistance device.
Conclusion: the future lies in collaboration
Beyond simple technological evolution, agentic engineering marks a structural transformation. Those who integrate AI agents into their toolchains today gain operational efficiency and, more importantly, lay the foundations for scalable, resilient and sustainable development processes.
The next generation of digital product development relies on close collaboration between humans and AI. It combines the best of both worlds: on the one hand, the precision of machines and, on the other hand, the creativity and responsibility of men. Here is the real breakthrough, that of engineering thought of as a field of authentic innovation.
