AI agents have rapidly advanced in software, a domain that has a clear iteration loop: write, run, verify, refine. The next frontier for AI agents is in long-horizon workflows. Domains like 3D and CAD sit on top of millions of expert hours that remain entirely manual to date.

Expertise in these fields live inside experts heads, across a stack of specialized tools. For AI agents to work reliably here, they will need data that captures exactly why and how work happens – not just final artifacts or recordings.

Shader is building the data foundry for agentic expert work. We capture native tool use traces from vetted experts performing production tasks inside specialized software – starting with 3D. This will lay the foundation needed to ship intelligent copilots and agents.