OriginShape joins NVIDIA Inception

The news

We’re excited to share that OriginShape has been accepted into NVIDIA Inception, NVIDIA’s program for startups. We build Oscar, the AI copilot for reverse engineering: its AI agents read a 3D scan like an engineer and rebuild the part as parametric CAD that you can edit and send to manufacturing.

Joining Inception gives a small team like ours access to NVIDIA’s training, tools and network while we make Oscar faster and more accurate on real parts.

What NVIDIA Inception is

NVIDIA Inception is a program that helps startups build and grow. It’s free: NVIDIA charges no application or membership fees and takes no equity. So NVIDIA doesn’t own any part of OriginShape, and Inception isn’t an investment. It’s a program that we’re a member of.

What the program gives startups

On its startup page, NVIDIA lists what members can get:

• Training. Free self-paced technical courses and discounted workshops led by experts, and access to the NVIDIA Developer Forums.
• Tools and offers. NVIDIA’s SDKs and model libraries, preferred pricing on select NVIDIA hardware and software, and cloud credits from NVIDIA and its partners.
• Investor introductions. Exposure to NVIDIA’s venture capital network, for startups that are eligible.
• Reach. Official badges, co-branded content, networking, and go-to-market opportunities.

The details vary by region and change over time. NVIDIA’s page has the current list.

Why it matters for reverse engineering

Reverse engineering a part is a lot of computing. A scan can hold millions of points. Oscar divides that surface into regions, fits planes, cylinders and other shapes to them, rebuilds the part feature by feature, and then compares the model with every point of the scan to show where it deviates. Its AI agents make each of those decisions the way an engineer does, one step at a time.

That’s the kind of work where GPU computing and good AI tooling help most. Through the program, we can learn NVIDIA’s tools, test them on our own geometry and AI work, and talk with NVIDIA’s engineers about what we find.

What it means for you

If you rebuild parts with Oscar, here’s what we’re aiming for with the program’s help:

• Faster results on big scans. Less waiting between uploading a scan and checking the model.
• Better checks. Deviation maps and reports that keep up with dense scans, so you can trust what you send to manufacturing.
• More parts that Oscar can rebuild. Our engine grows one family of parts at a time, and the training and tools help us move faster.

These are plans, not promises with dates. We’ll write about each change here when it ships.

What doesn’t change

• How you use Oscar. Oscar still runs in your browser and in FreeCAD, with the same account.
• What it costs. You can still start free and move to a monthly plan when you need more credits.
• Your data. We do not use your scans, files, prompts, or chats to train or fine-tune AI models. Our AI model provider processes your requests to write Oscar’s replies, and it does not train its models on them. Joining Inception doesn’t give NVIDIA access to your files.

Have a part without usable CAD? Try Oscar on it, or send us the part.

Links

• NVIDIA Inception. NVIDIA’s page for the program: who can join and what members get.
• Try Oscar. Oscar in your browser. Start free.
• How reverse engineering works. Our post on the seven steps, from the brief to delivered CAD.
• Oscar for FreeCAD is out. The open-source workbench that builds native FreeCAD features.

NVIDIA and NVIDIA Inception are trademarks of NVIDIA Corporation.

Questions