Zach, co-founder of Adam, recently shared that the company has moved beyond early text-to-CAD experiments to deliver a beta version of its AI CAD harness—one that integrates directly into industry-standard tools like Fusion 360 and Onshape. Earlier community feedback made it clear: engineers want AI assistance that respects their existing workflows rather than black-box outputs. Adam’s new approach delivers exactly that.
The harness reads your current part files, interprets the feature tree, and applies agentic edits while keeping all design decisions visible and editable. Early users are leveraging it for tasks like merging redundant features, bulk-renaming tree entries, applying consistent fillets, and even end-to-end parametric generation. Installation is now available through dedicated app store listings for both CAD platforms.
How Adam differs from earlier AI CAD tools
Adam’s approach prioritizes practicality over novelty. Unlike earlier text-to-CAD demos that generated standalone 3D models, this harness operates inside the CAD environment where engineers already work. It doesn’t replace your feature tree—it refines it.
The team emphasizes three core principles:
- Model-agnostic operation: Adam dynamically selects the best frontier model for each task based on internal benchmarks, rather than locking into a single vendor’s offering.
- Direct integration: Native CAD plugins for Fusion and Onshape ensure edits happen in context, preserving design intent and avoiding export/import friction.
- Engineering-first workflows: Tasks are framed around real CAD pain points—tree cleanup, naming conventions, parametric updates—rather than abstract “AI-generated shapes.”
Under the hood, the system relies heavily on Fusion’s Python API and Onshape’s FeatureScript, reinforcing the company’s long-standing belief in “CAD as code” as the right abstraction layer. Adam also maintains internal benchmarks to track spatial reasoning progress across frontier models, noting significant improvements in recent releases.
What engineers are using it for today
Early adopters aren’t just experimenting with AI—they’re solving concrete problems. Common use cases include:
- Consolidating duplicate or redundant features to simplify the feature tree
- Standardizing naming conventions across hundreds of model features
- Applying uniform edge fillets to internal geometry in bulk
- Introducing parametric variables to non-parametric legacy designs
- Generating new geometry from high-level instructions while retaining control
The team shared examples of engineers using the tool to clean up historical models, standardize design libraries, and even explore new parametric workflows—all without leaving their CAD environment.
Looking ahead: open development and industry validation
Adam has open-sourced earlier text-to-CAD prototypes, signaling a commitment to transparency. The company also acknowledges recent industry moves, such as an Anthropic-Autodesk connector, as validation of the in-CAD AI direction. Where Adam stands apart is in its model-agnostic architecture and deep integration philosophy.
The team is actively expanding platform support and refining task-specific benchmarks. They’re also inviting feedback from mechanical engineers: What in-CAD automation feels missing today?
As AI capabilities in spatial reasoning continue to evolve rapidly, tools like Adam could redefine how engineers interact with CAD—not by replacing them, but by removing repetitive friction so they can focus on solving the real problems in front of them.
AI summary
Adam adlı AI destekli CAD aracı, Onshape ve Autodesk Fusion entegrasyonlarıyla mühendislerin özellik ağaçlarını optimize etmelerine yardımcı oluyor. Yapay zekanın CAD alanındaki geleceği hakkında detaylar.
