I joined this team a few months ago, and honestly it's been one of the craziest stretches of my career.
Rewind three or four years before this AI era really hit, writing code was the hard part. Slow, painful, every feature a grind. You measured progress in weeks.
Now I watch how in Autonomyware three/four words and it comes back with the entire product: the requirements, the 3D model, the risk analysis, the whole chain. The first time I saw it run end to end, I just kind of sat there. The speed still doesn't feel real, even building it from the inside.
How does yhe system handle changes when one part of the design affects everything else?
Autonomyware
@advin_jadis thank you. Our tool creates complete products as one CAD program. Every verified part created carries a fingerprint of the exact program slice its geometry depends on (its construction, placement frame, shared datums, helpers and parameters). When a part is changed, only that part and the parts whose slice runs through the changed lines are rebuilt; every other part keeps an identical fingerprint, is frozen, and its exact verified body is restored untouched — and a patch that would silently alter a frozen part is refused unless a finding names it, so ripple effects are explicit. After the rebuild the whole assembly is re-verified (mates, interference, floating, completeness) and every finding is handed back to the author with a repair, while the traceability cascade re-marks the affected BOM, interfaces
What types of tests or checks can the AI prepare before a product reaches production?
Autonomyware
@austinphillips great question!
We have 2 operating modes:
1. Forge - physical product layer (exact b-rep CAD, ECAD, mesh). The checks here are assembly, fitting, clash, motion, wall thickness and visual coherence oriented.
2. Autonomous engineering - product definition > design decomposition > formal verification, validation and unit test creation > realization for all disciplines. The tests here are actual engineering V&Vs (system integration testing, subsystem testing, component and unit testing), traceability checks and second party reviews (the main engineering model is reviewed by a second model).
I have been manually managing hardware risk logs and BOMs across separate spreadsheets so having an integrated workspace for every step saves so much time.
Autonomyware
@jack_hayes83 - absolutely! Not to mention the risk of rework and reengineering that comes from gaps in traceability coverage or not being able to understand how a part change impacts the risk analysis. Every engineered prototype product run in Autonomyware is connected end-to-end in traceability - from assumptions, to specs, operating contexts, design, BOM and verification.
Can it work with different manufacturing methods such as CNC machining injection molding and 3D printing?
Autonomyware
@linpeng That’s a great question! We provide different modes focused on different outcomes, for example rapid prototyping and 3D printing (e.g. GLB files) or traditional manufacturing like CNC and also injection molding (oftentimes B-rep data). For complex products, Autonomyware provides manufacturing instructions. You can find different examples in our product library. These were created during our Alpha Testing program! https://autonomyware.ai/library
I like the idea of not needing to learn CAD first just to see if an idea is worth making.
Autonomyware
@kyle_bennett6 - thank you! Exactly, from natural language to fast prototyped design regardless of technical knowledge. Like vibe-coding, but for physical products.
It would be nice if rotating an object in the viewer worked without such annoying inertia. Otherwise, it's a great product.
Autonomyware
@maxim100000 Thanks for your good and honest feedback. We will tweak the control of the viewer in our next version. Glad that you enjoy working with Autonomyware.
Tested a concept prompt this afternoon it mapped out the architecture, code and BOM layout in just a few mints.
Autonomyware
@andrew_dale2 - thank you, Andrew! From idea to a fast, engineered prototype ⚡️