Citation :

Abirami, Swasti Karna, 2025. "AI-Native Engineering Workflows: Embedding Generative Models into System-Level Design for 2025" ESP International Journal of Emerging Multidisciplinary Research [ESP-IJEMR]  Volume 1, Issue 1: 53-63.

Abstract :

The exponential increase of generative artificial intelligence (AI) has introduced a new paradigm in the way we conceive, model and validate engineering systems. AI technologies are commonly added as helper appendices to good old computer-aided design (CAD) and system engineering workflows, which remain largely deterministic automated and manual despite strong organisation and rich in tools. To bring generative design models, such as large language models (LLMs), diffusion networks, graph-based generators and domain-specific neural optimisers into the infrastructure level design life-cycle, we introduce the concept of AI-native engineering workflows: a next-generation approach. In contrast to traditional “AI-assisted” methods, AI-native workflows embed generative models at all stages of the engineering orchestration pipeline: requirement definition and conceptual synthesis through validation, deployment, and lifecycle monitoring. To ensure reliability and trustworthiness, the proposed system strongly emphasizes a human-in-the-loop governance model, witnessed provenance, and closed-loop validation. To support scalable, secure and transparent integration of generative technologies into existing engineering ecosystems, we propose an architectural model with five functional layers – where the former include: interface, orchestration, validation, provenance and deployment. We demonstrate the measurable benefits of AI-native workflows with twofacilitated `what-if´ case studies: a distributed softwaresystem control plane, and a mechatronic subsystem ConceptualDesign. These benefits are: improved decision support by means of natural language interfaces, improved variety on the design process, automatic artefact generation and faster prototyping. Quantitative measures are also provided to assess safety compliance, explainability, design fidelity and productivity. Results demonstrate that AI-native workflows can help you validate more solutions, automate the documentation and testing process without a decrease in quality, accelerate early-stage design cycles by 40–60%. However, these advantages are counterbalanced by shortcomings in terms of ethical control, infrastructure quantity, repeatability and verification. The research concludes that robust validation pipelines, model versioning and collaboration between software engineers, data scientists and domain experts are key to the successful deployment of AI-native workflows. A key milestone towards a symbiotic human–machine co-design paradigm that will define industrial engineering practices post 2025 is finally provided by the work in this study, which sets the concept and methodological underpinnings of AI-native systems engineering.

References :

[1] Agrawal, A., & Verschueren, R. (2024). Differentiable physics and its role in generative engineering design. IEEE Transactions on Industrial Informatics, 20(2), 1021–1034.

[2] Anderson, P. (2023). Integrating AI into systems engineering: A survey of emerging practices. Systems Engineering Journal, 26(3), 275–289.

[3] Banerjee, R., & Kumar, S. (2024). AI-native architectures for digital twins. Journal of Computing and Information Science in Engineering, 24(4), 041002.

[4] Bell, D., & Zhao, L. (2025). Generative AI in mechatronic system design: Emerging frameworks. Advanced Engineering Informatics, 55, 101654.

[5] Boström, E., & Liang, Y. (2024). Embedding LLMs in engineering workflows: A practical perspective. AI and Society, 39(5), 1133–1150.

[6] Brown, T. et al. (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems, 33, 1877–1901.

[7] Chen, J., & Gupta, P. (2023). Autonomous co-design using simulation-backed generative networks. Journal of Mechanical Design, 145(6), 061701.

[8] Clark, D., & Regev, A. (2024). Governance challenges in AI-driven engineering. Engineering Management Review, 52(2), 65–78.

[9] Das, N., & Yu, C. (2025). Closed-loop verification in AI-native control systems. IEEE Transactions on Automation Science and Engineering, 22(1), 50–63.

[10] Deng, W., & Liu, T. (2023). Traceability and provenance in generative pipelines. Computers in Industry, 150, 103845.

[11] Evans, J. (2024). Digital twins and generative feedback loops. Smart Systems Review, 18(3), 199–214.

[12] Fischer, B. (2023). Generative AI in software engineering: Opportunities and risks. Empirical Software Engineering, 28(5), 86–103.

[13] Gao, X., & Petrov, I. (2024). Neural compilers for autonomous code synthesis. ACM Transactions on Software Engineering and Methodology, 33(2), 34–49.

[14] Gonzalez, L. (2025). Adaptive manufacturing through AI-native design loops. Journal of Manufacturing Systems, 76, 210–225.

[15] Huang, C., & Meyer, J. (2024). Graph neural networks in topology optimization. Engineering with Computers, 40(2), 357–369.

[16] Johnson, R. (2024). Ethical AI governance in autonomous engineering systems. AI Ethics Review, 9(1), 41–58.

[17] Kim, H., & Lee, D. (2023). Validation-aware generative models for structural systems. Computer-Aided Design, 157, 103567.

[18] Kiran, S. (2025). Human-AI collaboration in mechanical engineering design. International Journal of Advanced Engineering Research, 12(1), 45–60.

[19] Lin, J., & Fang, Z. (2023). Hybrid generative-surrogate modeling in aerospace design. AIAA Journal, 61(9), 3743–3758.

[20] Liu, X., & Singh, V. (2025). System-level integration of generative models in control systems engineering. Control Engineering Practice, 145, 105984.

[21] Ma, Y., & Krishnan, P. (2024). Model provenance tracking for explainable AI pipelines. Journal of Data Intelligence, 6(3), 245–262.

[22] Nakamura, T. (2023). Simulation-informed AI design in robotics. Robotics and Autonomous Systems, 165, 104361.

[23] O’Reilly, P., & Wang, Y. (2025). AI-native digital ecosystems for smart manufacturing. Computers & Industrial Engineering, 190, 109829.

[24] Patel, R., & Zhang, W. (2024). Mechatronic co-design using generative agents. Mechatronics, 96, 103254.

[25] Perez, M., & Zhou, J. (2023). Template-driven generative frameworks for cyber-physical systems. Future Generation Computer Systems, 150, 412–427.

[26] Rajan, N. (2025). Generative AI in embedded system verification. Microprocessors and Microsystems, 105, 105357.

[27] Sharma, A., & Kim, S. (2024). AI orchestration in multidisciplinary design environments. Journal of Systems and Software, 205, 111939.

[28] Singh, R., & Thomas, E. (2025). Performance metrics for AI-native engineering workflows. Engineering Applications of Artificial Intelligence, 135, 108384.

[29] Tanaka, H., & Li, X. (2024). Generative validation pipelines in hybrid AI systems. IEEE Access, 12, 123876–123889.

[30] Zhao, K., & Park, M. (2023). Reproducibility and ethics in generative design. Design Studies, 89, 102128.

Keywords :

AI-Native Engineering, Generative Design, Large Language Models, System-Level Design, Workflow Orchestration, Validation Frameworks, Human-In-The-Loop, Traceability, Simulation-Driven Design, Intelligent Automation.