neThing.xyz

• 

  Screenshot showing the generation of a pipe with flange fittings

Inspiration

Our mission at polySpectra is to help engineers make their ideas real. We have been incredibly energized by the pace of innovation in generative AI, but we were unable to find a text-to-3d gen AI tool that was able to create objects that we would actually want to manufacture.

The key insight that led to the invention of neThing.xyz was that AI is actually quite good at writing code, and by training our "code gen" AI on domain specific languages, our AI can produce code that can be rapidly converted into a 3D CAD models.

What it does

1. It’s pronounced “any thing dot x,y,z”.

2. It is a coding-based approach to text-to-3D generative AI, broadly based in CAD. Most of the other apps in this space are using NeRFs, which is basically a combination of text-to-image and then 2D-to-3D. (Lumalabs “Genie” and Commonsense Machines “Cube” are our two favorites.)

3. You can prompt the AI to generate the code for the 3D model, which in turn generates the 3D model. You can preview the file in AR (on compatible devices), change the physical appearance, edit the code yourself, download the corresponding .STL or .GLB files for 3D printing or further 3D modeling, or have the polySpectra team additively manufacture the part for you with the “Make it real” button.

How we built it

AI wrote most of the code for the app (Cursor.sh is our most productive team member). For this RAG-a-thon, the key focus was on adding a RAG pipeline to the app, which we achieved with Llamaindex and AstraDB. We are particularly grateful to Logan from Llamaindex for his help, as well as the talented and generous community behind Build123D

Challenges we ran into

Llamaindex and AstraDB have some homework to do, the documentation is great for setting up the initial vector store, but there is no documentation on how to use it in production. The Llamaindex Docstore Pipeline workflow currently doesn't work with AstraDB. Please help!

Adding Llamaindex broke our current observability platform, so we will need to set up something new.

Accomplishments that we're proud of

Using our Llamaindex and AstraDB RAG workflow, we were able to achieve a 5x reduction in the number of tokens required to get the AI to generate functional 3D models. We do not have a big budget, so this 80% reduction in our OpenAI bill per user query is incredibly meaningful to us.

What we learned

Dream big, but start with very small goals. Everything takes longer than anticipated.

Ask for help! We're working on an incredibly challenging problem and we can't solve it alone.

Share your wins! It was really fun to see everyone's Day 1 accomplishments. I can't wait to see the pitches today.

What's next for neThing.xyz

Help engineers make their ideas real by creating the best text-to-CAD generative AI out there!

(There are simply too many ideas to list here.)

Check out the tool : https://neThing.xyz Join our community: https://forum.neThing.xyz

Built With

• astradb
• datastax
• llamaindex