Open-Sourcing 99 Hours of CAD Workflows to Train AI Engineering Agents
DevvMandal · x · 2026-08-04
Markov AI has released an open-source multimodal dataset of CAD environments designed to train AI models for engineering tasks.
- Scale: Contains 51 task workflows totaling 99.03 hours of human walkthrough screen recordings.
- Software Coverage: Spans 8 major mechanical design and architecture applications, including AutoCAD, SOLIDWORKS, and Siemens NX.
- Rich Context: Each workflow preserves the full context—not just the final model—featuring task prompts, reference files, gold outputs, evaluation rubrics, synchronized interaction events, and narration.
This provides a high-quality baseline for evaluating and advancing autonomous AI agents in complex desktop software environments.
Related event: Markov AI Open-Sources Multimodal CAD Environment Dataset(3 posts)→
More from coding & agent
- Scalar Field Launches Agentic Trading Desk Connecting Research to Live Execution — ycombinator · 2026-08-04
- Vibe Coding in Virtual Reality on the Apple Vision Pro — nptacek · 2026-08-04
- Can AI Coding Agents Crack Nvidia's CUDA Moat? — bingxu_ · 2026-08-04
- HeyGen LiveAvatar Agent Builds $3M Pipeline During 8-Week Employee Leave — HeyGen · 2026-08-04
- Agent Benchmark Reflections: Scores Are Deceptive, Open-source Trajectories Needed — Shahules786 · 2026-08-04
- Agent Benchmark Flaws: Over-specified Verifiers Penalize Semantically Correct Actions — Shahules786 · 2026-08-04