Independent company profile · AI for engineering

Zoo: Zookeeper in Zoo Design Studio for Engineering — What Mechanical Engineers Should Know

A mechanical-engineering look at Zoo Zookeeper, a conversational CAD agent built around parametric B-rep modeling and a code-capable CAD engine.

BM
Written by Bertrand Mezatio

Mechanical engineer focused on CAD, DFM and manufacturing. This profile is independent editorial analysis based on public company information.

Independent editorial profile: DIVE-LD is not affiliated with, sponsored by or endorsed by Zoo. Product capabilities change quickly; verify current functionality with the vendor before making engineering or procurement decisions.
Engineering categoryConversational parametric CAD
Workflow stageCAD creation, editing, review and agentic design
Typical inputNatural-language instructions, selections and existing CAD context
Typical outputEditable parametric B-rep CAD inside Zoo Design Studio

What Zoo is building

Zoo's Zookeeper is a conversational CAD agent integrated into Zoo Design Studio. The company describes it as able to create new models, modify existing designs, inspect selected faces/features, perform research and iterate through geometry while remaining connected to Zoo's CAD engine. Zoo also exposes a readable CAD language and engine-level tools, giving the agent a structured way to create and validate parametric B-rep geometry.

Where it sits in the engineering workflow

The key architectural idea is tight integration between the agent and the geometry kernel. A language model can describe a fillet, but useful CAD automation requires the system to identify the correct edge, apply the operation, detect whether the model remains valid and recover when the kernel rejects the feature. Zookeeper's tool-using loop is therefore more relevant to engineering than a chat window placed beside a CAD viewport.

Why mechanical engineers should care

Conversational editing may reduce repetitive interaction and make CAD more accessible, while selection awareness can keep prompts grounded in the actual model. For experienced users, a productive agent should behave less like a replacement designer and more like a programmable junior operator: carry out well-defined operations, explain what it changed and let the engineer inspect the resulting feature history.

What it does not remove from engineering

Agentic CAD can fail in ways that are subtle: wrong reference selection, fragile feature order, geometrically valid but functionally wrong features, or an operation that succeeds on one configuration and fails on another. A professional workflow needs regeneration tests, model checks and downstream DFM/CAE validation before release.

Questions to ask before adopting it

  • Can the agent explain exactly which faces, edges and parameters it changed?
  • How does it recover from failed kernel operations?
  • Does the feature history remain readable and stable after repeated conversational edits?
  • Can engineering teams version and review agent-made changes like normal CAD changes?

DIVE-LD engineering view

The most useful way to evaluate Zookeeper in Zoo Design Studio is to put it inside a controlled engineering release process. Ask whether the output preserves design intent, whether assumptions are visible, whether another engineer can edit or audit the result, and what verification remains before manufacturing. AI can compress repetitive work and widen the design space, but responsibility for requirements, safety, standards, tolerances and final release still belongs to the engineering organization.

Sources and further reading

Public product information reviewed on 11 September 2026. Company statements about performance or capability are vendor claims unless independently verified.