Ai Build introduces GPT for additive manufacturing

For a company whose name is Ai Build, the idea of using the latest advanced artificial intelligence chatting software should not come as a surprise. The London-based AM software developer has now introduced “Talk to AiSync” – a natural language-based system for giving a 3D printer slicing instructions that is currently entering the beta testing stage with selected partners, before being released more generally.

“We are working with the industry leaders to establish the boundaries (if any) of this breakthrough development in additive manufacturing workflow automation,” said Daghan Cam, Co-Founder and CEO of Ai Build, in a LinkedIn post. “The underlying technology is the combination of Ai Build’s visual toolpath programming engine and OpenAI’s large language model GPT, fed with years of real-world manufacturing data from our AiLab in London.”

According to Daghan, this is not a replacement for designers or engineers, and the human is still very much in the loop for decision-making. It is, however, a new kind of natural language interface that elevates the user experience.

For example, to create a highly optimized 3D printing file for the popular Stanford Bunny, Ai Build used prompts such as:

1. “Slice the bunny with 2mm layer height.”
2. “Increase the infill gradually in the last few layers to make the ears stronger.”
3. “Printing speed looks good overall but reduce it a little bit on sharp corners to maintain better accuracy.”

“Talk to AiSync” is expected to massively reduce the barriers to entry for the mainstream adoption of AM. Anyone that can type or say “Print my part” will be able to print their part successfully without the guesswork, and trial and error. Ai Build measures this impact with a First-Time-Right (FTR) ratio.

The other anticipated impact of Ai Build’s GPT is that it will empower designers and engineering teams to optimize production at ‘superhuman levels at lightning speed’. Teams will now be able to focus on crafting complex multi-objective engineering intents and creating better products made without the limitations of hard-to-master graphical software interfaces. “Print my part nonplanar without using support, minimize start-stops, and make sure there is at least 2mm additional stock on the top side for CNC milling afterwards,” is the kind of prompt that can be given to achieve application-specific requirements. Ai Build measures this impact as the average Design-To-Production-Time that is spent in front of the computer for preparing fabrication files.

As interesting and exciting as this new development is, one can only imagine the future impact artificial intelligence is going to have on 3D printing, and manufacturing more broadly.