Adam, a new AI copilot, plugs directly into Autodesk Fusion 360 and Onshape, letting engineers edit feature trees with agentic precision. Zach, a co‑founder, notes that mechanical designers reject black‑box STL generators for full visibility inside their CAD environment. The extension reads existing models and rewrites features on demand in real‑time sessions, supporting advanced automation and efficiency.

Behind Adam lies a benchmark that pitted frontier language models against CAD tasks. Results show a leap in spatial reasoning, with GPT 5.5 and Opus 4.7 outperforming earlier releases. The team switches models per task, ensuring the most accurate tool handles feature recognition, parametric changes, and edge filleting by leveraging deep learning architectures within the plugin framework daily.

Users report practical gains: merging redundant features, renaming trees for readability, adding uniform 2 mm fillets, and fully parametrizing models. Adam also supports end‑to‑end CAD generation, a feature that attracts designers seeking rapid prototyping. The extension installs via a simple script, embedding itself in Fusion’s Add‑Ins folder, and activates after a single run every session without.

By staying model‑agnostic, Adam selects the best frontier model per task rather than locking into a single vendor. Open‑source roots trace back to earlier text‑to‑CAD experiments, and the team welcomes community contributions. In practice, the copilot streamlines routine CAD edits, freeing engineers to focus on design innovation and maintain quality standards for industry applications.