After an earlier story today on how AI could penetrate the AM space, a company announced exactly that.
Ivaldi Group announced an unnamed ChatGPT integration that can assist their customers in several important ways.
ChatGPT is OpenAI’s increasingly popular text AI tool that can provide astonishingly good results in many areas (just don’t try any mathematics). A brief text input is seemingly understood by the AI, and it rapidly provides an appropriate result.
OpenAI has enabled an API for ChatGPT, which permits anyone (with appropriate credentials) to build apps that call upon ChatGPT in the background. That’s what Ivaldi Group has done here.
If you’re not familiar with Ivaldi Group, they are a distributed manufacturing service that focuses on heavy industry, specifically marine and mining applications. With marine applications, for example, it’s possible for a ship to order a spare part to be 3D printed at the next port of call for near-immediate installation. That on-demand capability can short-circuit expensive delays and make shippers far more efficient.
What exactly does this integration do? Ivaldi Group explains:
“Ivaldi’s custom ChatGPT integration allows secure equipment data and maintenance documentation to be indexed and accessed through an AI assistant. Maintenance crews can engage in custom dialogues with the AI system for tailored information, eliminating the need for physical manuals or online document searches.”
Ok, that’s good. Thick paper manuals are never a good thing. But where does 3D printing come in? Ivaldi continues:
“The platform also facilitates ordering replacement components directly through the interface. If a part is not already in the Ivaldi virtual library, users can submit 2D drawings or simple sketches, which the system converts into custom-generated models within pre-engineered categories. Once approved, Ivaldi then creates a printable 3D model and coordinates production and delivery with its partner network of 15,000 local manufacturers.”
This sounds a lot like the “text to 3D” systems that are being developed by Luma Labs, NVIDIA, OpenAI, and no doubt many others.
However, the difference here could be that Ivaldi Group is focusing the results down to “pre-engineered categories”, which could mean an existing library, or it could mean that the AI is trained to generate 3D models that are similar to the geometries of typical parts in that industrial domain.
Regardless of which approach is used, the system appears quite powerful. Check out this example where a sketch has been made on a dirty car window, and then transformed into a real, 3D part.
[Aside: 2007-Me would be quite puzzled after reading that last paragraph.]
Ivaldi Group’s approach here seems unique — at least for now. It would seem reasonable that competing manufacturing services might attempt to build their own AI interfaces to provide similar services.
The new AI tech is opening up all kinds of possibilities, and today we’ve seen but one of many to come.
Via Ivaldi Group
