What is Generative Engineering?
What is Generative Engineering?
UK Software Startup Showcasing Their Capabilities at CDFAM NYC 2024
Can you introduce us to Generative Engineering, how your backgrounds have influenced the direction of the company and what it is you offer as a software or service?
As founders we came together in a startup engineering company trying to redefine how electric vehicles were designed and produced. We experienced first hand how the patchwork of engineering software tools available were making it hard to develop products leveraging the latest technologies at speed.
We resolved to fix this problem at its root by bringing together our unique set of skills and experiences from decades working at the cutting edge of software development, cloud computing, AI, and computational design.
The result is Generative Engineering, a software platform for radically improving the engineering design process. It integrates into and complements existing engineering software tools, both locally and in the cloud. This includes CAD and simulation tools.
Our product generates and analyzes thousands of design possibilities at once, ensuring teams of engineers can rapidly collaborate on data-driven design insight with AI suggestions and automation.
Our aim is to give engineers all the data, information, and hindsight they obtain by the end of an engineering process, at the very beginning. We want to establish effective decision-making throughout every stage of engineering, by showcasing and analyzing all possible design directions, combined with a clear explanation of the possible decisions to all involved, whilst complementing all other tools in the engineering stack.
Your presentation at CDFAM NYC aims to address the gap between AI advancements, and their application in engineering and high-value manufacturing. What do you see as the primary obstacles to integrating AI more fully in these fields?
The bar for AI in engineering has to be set high. Engineering is a precise discipline that requires reliability and explainability in the outputs of any process. But there are less precise things in the process that AI can help with, given key barriers are overcome:
Trust: New tools have to be trustworthy and black box machine learning algorithms dont always explain how they come to a particular output.
Difficulty of automating engineering software:
Engineering is a relatively data poor industry because engineering data is expensive to create, multi-dimensional and multi-modal, and often siloed.
Data quality usually suffers from limited versioning, data syncing issues, and missing data.
Usability has long been an issue for engineering tools in general and is a particular barrier for more advanced computational methods which require learning new technical skills such as coding or PhD level optimisation techniques.
What specific strategies or technologies is your team working on to overcome this?
We help engineers generate large data sets to help them understand engineering problems, and we use AI to assist them in reaching the insight needed to make effective design decisions from this data.
The Generative platform makes the barrier to doing so lower than ever before. It provides automations on top of tools that engineers already know and trust: traditional CAD & numeric simulation tools (and we employ AI assistance to make it quick and easy to set up new designs.
How can engineering teams build trust in AI-driven tools and processes, especially given the challenges of “black box” systems like LLMs and VLMs, which sometimes produce results that don’t align with the harsh realities of physics?
The idea is to work together with AI, rather than expect it to come up with all the answers for you. When engineers leverage it to guide their decision making, they can accelerate their understanding of the problem and ultimately make more informed decisions.
We combine AI suggestions with rigorous computational engineering to overcome some limitations of black-box AI systems. This lets engineers reach the right simulations and design ideas quickly, showing promising potential even with today’s technology.
What data inputs and outputs are used in your process, what does a client need to prepare in terms of data and how does this influence the outcomes of your AI-driven engineering solutions?
We start with the CAD and numeric simulation tools that engineers already use today. If you’re creating a BRep CAD, and are running it through a simulation, we make it easy to move to automatically generating variations that answer engineering questions.
The more data you provide the better – whether it’s documents that LLMs can interpret to make suggestions, or simulation data you’ve already run. But we make sure you can start from whatever data you already have available, and get to the data needed quickly.
What key insights or lessons do you hope the audience will take away from your presentation?
AI will not take your job. But AI-assisted engineers will outcompete non AI-assisted engineers.
That leveraging generative techniques and AI assistance is no longer reserved for those with specialist skills.
There is a change coming in the industry, and now is a good time to get on board with the change and help others through it.
Finally, what are you looking forward to gaining from attending CDFAM NYC?
To have a great time. Also to meet many other people who are also trying to make this change happen, to get feedback on early versions of what we’re building, and to find people interested in trying it out.
