Engineering: AI’s last frontier?​

• Engineering is at a turning point. As AIs automate increasingly complex tasks, most of us are wondering what will remain human, in an AI-driven engineering world.
• In this article, we explore how engineering teams are moving beyond simple automation to embrace collaboration with AI co-pilots that operate alongside them.  
• The recent integration of Neural Concept with Nvidia Omniverse was built around the symbiotic human-AI interplay, which takes life out in real-world cases like SP80’s pursuit of the sailing speed record. ‍
• Pierre, Neural Concept’s founder shares his vision of what is ahead of us.

Why AI is Rewriting the Engineering Playbook

Engineering is undergoing one of its most significant transformations. The pressure on advanced manufacturing has intensified in recent years. Companies face massive product evolutions while striving to increase manufacturing speed and reduce costs. At the same time, a massive shift in engineering skills towards Machine-Learning and the progress of AI tools are providing new ways to tackle challenges.

Across all sectors, an AI-driven engineering transformation is a matter of survival. Companies are racing to embed AI into engineering processes, transforming into AI-driven hardware design leaders. Yet, as AI takes on more design tasks, a critical question arises: what will be left for human engineers within this new paradigm?

We believe the answer lies in a symbiotic collaboration between human creativity and AI’s analytic power. ​Our Neural Concept platform marks a major step in that direction, leveraging the NVIDIA Omniverse platform to create​​ a dynamic, interactive environment where engineers and AI can work together to push the boundaries of product design.

{{highlight}}

AI is not just accelerating product design—it is reshaping how engineers explore design spaces, evaluate trade-offs and it enhances decision-making at every step.

Since engineering performance cannot be expressed via equations only, human feeling and expertise are paramount.
‍

How Human-AI Collaboration Really Works

A powerful AI-Human collaboration workflow will typically unfold in the following way:

• An AI Engineering Expert sets up an ​Engineering ​AI co-pilot for generative design on a specific class of challenges in the context of its own company. For instance, aiming at designing the high-pressure stage of a gaz turbine.

• A designer prompts the AI system with specific design targets, while specifying manufacturing and design constraints.

• The Engineering AI Copilot generates thousands of design variations optimized for physics-based criteria.

• The human designer evaluates and refines designs visually, incorporating human preferences, expertise or unspoken constraints.

• Human preferences are fed-back, parameters, adjusted, and the AI starts re-generating designs based on feedback.

​​     ​​This continuous loop between machine intelligence and human feedback unlocks smarter, faster, and more impactful design decisions.​
‍

Real-World Application

This is not just a theoretical framework. Teams today are already leveraging AI-human collaboration to push the boundaries of physics and design.

A compelling example of this approach is SP80, a Swiss engineering team aiming to break the world record for wind-powered speed. The SP80 team ran hundreds of thousands of design optimizations with Neural Concept, leveraging its power​ to explore 10,000+ design possibilities. However, despite the optimized hydrofoil meeting all computational requirements, once tested on the water, the pilots sensed that the boat’s behavior was not optimal.

This led to a key insight: AI-generated designs need human judgement.​ With Neural Concept’s recent addition, the SP80 team was able to leverage the capabilities of the ​​NVIDIA Omniverse​​ platform to build a hyper-realistic simulation environment where they could visualize and interact with thousands of design options before committing to a final one.​ This iterative, human-in-the-loop approach allowed them to refine their design based not just on physics-based optimization but also on human judgment, accelerating qualitative feedback from non-technical stakeholders like the pilot.

Breaking World Records with AI & Simulation

• Targeting a world record-breaking 80 knots (~150 km/h) in 2025, aiming to surpass the previous world record by 18%.

• Pushing the boundaries of physics, the team designed a hyperventilating hydrofoil to mitigate cavitation—an extreme phenomenon that risks structural collapse and severe instability.  

• The previous record holder faced a near-fatal crash in their first attempt. SP80 is balancing the pursuit of record-breaking speed with a safety-first approach, ensuring performance does not come at the cost of stability.

• Leveraged Neural Concept & NVIDIA Omniverse to explore 10000+ times more designs, rapidly incorporating pilot feedback and refining boat behavior in extreme conditions.

The same capabilities that enabled SP80 to push the boundaries of engineering also apply to more traditional industries. In automotive, aerospace, and consumer product development, AI-driven workflows enable teams to collaborate 10X faster on parts and system designs – by integrating physics, simulation, and real-world expertise.

‍

‍

<div id="founders-vision"></div>

​​​​​​​Founder’s Vision: Why Engineering is AI’s Last Frontier

By Pierre Baqué, Founder and CEO at Neural Concept

AGI's last frontier might be engineering. But it will eventually be crossed.

​​Automation of Engineering with AI

​​​Every time I think about it, I am amazed. A machine, built by humans, that can take a full family across 1000 km without stopping, within a few hours, and ride 200,000 to 500,000 km with minimal repairs. A vehicle affordable for a price equivalent to a year’s average salary and yet providing a deep feeling of attention to details and comfort. All this, engineered almost from scratch and manufactured at scale in just a few years.​​

​​The more I have been interacting with automotive suppliers and manufacturers in the last 10 years, the more I have been struck by the subtlety, intricacy and complexity of human expertise, digital chains and interactions that allow for these repeated engineering feats. Although the process has been codified and optimized at its meta-level, and the design of individual components is mostly automated, the result very much emerges from a network of AIs, humans, and manufacturing machines. It will take a very long time of standardization and learning for AIs to progressively handle the entirety of this process. Replacing it from scratch with a single giant AI model starting from a blank page seems equally complicated, given the variety and complexity of constraints and physics to be considered. I am confident to say that Engineering may be one of the last domains where human decision-making remains essential for years to come. ​​

However, tremendous efficiency gains will be brought by AI into the design process in the coming years, and this is Neural Concept’s core mission. Technology will evolve towards replacing massive amounts of human intelligence in current traditional engineering works with AI.
‍

​​​What is left for humans?

​So, humans need not apply? Not quite. I believe that human jobs in engineering will evolve in two directions:​​

• ​​​Those building and training AIs​​
• ​​​Those expressing human preferences​​
  ‍

​​​The AI builder​​

​​​AI builders will not only be building AIs but also ensuring their trustworthiness, performance, and continuous improvement. Scripting, running systematic design studies and creating statistical analyses of physical experiments has long been part of the role of many engineers – particularly simulation engineers. In the world of AI, this role is taking up more space, becoming one of the focal attention points of many companies and the most decisive expertise for their competitiveness. As most engineering companies are moving from being hardware designers to builders of AIs, this is the key job of tomorrow. Neural Concept has been working with these engineers for years now and has developed the world’s best platform for these brains to create value and impact in their companies and 10x the usual speed and with maximum value added. We are launching Neural Concept Spark Sessions — a discovery trial for simulation engineers to implement AI-driven design workflows, assess their readiness to develop them, and learn data science best practices tailored for engineers.​​ ​​​​​
‍

​​​Expressing human tastes​​

​​​That’s not all. Most advanced consumer products are only convincing to humans when they achieve a subtle — and somewhat magical — alignment between technical precision and human taste. Therefore, the second class of workers will be the creators. The ones who are injecting the unquantifiable element of evolving feeling in designs. No more than I would like to make love with a perfect robot, would I like to buy a car, a boat or a phone that does not originate from a human soul. That’s why we are introducing the Omniverse integration, a new step towards harmonizing the cold AI-based generative design and the art and beauty of human experience and preferences. By enabling engineers to visualize and compare AI generated designs in a highly realistic environment, to compare concepts and express preferences, the new Omniverse integration creates this essential feedback loop.​​
‍

​​​SP80 and the power of a Sailor’s feeling ​​

​​​Most sailors will tell you that sailing is as much of an art as a science. The sailors’ feeling and intuition cannot be dissociated from their performance. Great ship designers don’t just calculate forces, pressure, and momentum; they envision the experience of sailing their creation. SP80 is a team of Swiss sailors and boat designers looking to break the world record of wind-powered speed. I follow them closely and I can tell you that they will do it. However, after running hundreds of thousands of multi-physics generative optimization campaigns with Neural Concept, the SP80 felt that something was missing. As they were testing the latest generation of their hyper-activating hydrofoil, they immediately felt on the water that the boat’s behavior was not right. It was not a calculation mistake. The foil was exactly optimized to match the equations they had entered. But their sailor’s feeling told them that the design was not right.​​

​​​They understood that to reach new levels of performance, they needed to combine the sheer power of AI-based generative design with their pilot’s intuition in a much tighter way. Hence, they asked us at Neural Concept to integrate Omniverse as a realistic visualization environment, combined with the online loop. We contacted NVIDIA, ran a client panel to assess needs and understood that this requirement was absolutely shared and that we would be missing an essential piece of our puzzle by not providing this closer human/AI integration. With this new approach, SP80 has refined its hydrofoil and is now actively testing in real-world conditions—combining AI’s computational power with a sailor’s instinct to achieve breakthrough performance.​​
‍

See Engineering AI in Action

At NVIDIA GTC 2025, we showcased how Neural Concept and NVIDIA Omniverse empower teams to bring AI into real-time design decisions:

• Teams ran live simulations and optimized SP80’s hydrofoil inside Neural Concept Studio and Omniverse.

• Engineers explored thousands of design variations in minutes, adjusting parameters and refining performance collaboratively, not sequentially.

• We shared how companies like General Motors are redefining safety and performance using AI-driven workflows.

Watch the Neural Concept & General Motors GTC replay to see how large-scale AI adoption is reshaping engineering’s toughest challenges.
‍

Build Your Own Engineering AI Co-Pilot

We’ve only scratched the surface of what’s possible when engineers and AI create together. If you’re ready to start this journey, we invite you to apply for a Neural Concept Spark Session:

A hands-on, guided discovery workshop to:

• Explore AI-driven design workflows tailored to your engineering challenges

• Build your first Engineering AI Co-Pilot on the Neural Concept platform

• Learn practical data science best practices for engineers

Whether you're designing the next record-breaking hydrofoil or optimizing an automotive part, Spark Sessions help you bring AI into your design loop—faster.

Learn more & Apply for Spark Sessions

AI will accelerate what machines do best. But the human role—creativity, intuition, judgment—will remain essential for years to come.

Join us in shaping this next era of AI-driven engineering.