Course 1 · Product Design · Beginner

How to use Blender — the viewport and the five moves

You cannot direct an agent in software you cannot operate yourself — if you do not know what "scale", "extrude", or "the origin" mean, you cannot tell the agent what went wrong or even see that it did. Before the first prompt, you need the moves.

Read this module through your lens

Designers: you may know other 3D tools — this is a fast map of Blender's verbs so the rest of the course makes sense.

Why a beginner starts here

This course is about directing AI agents, not about modelling by hand — so why touch the Blender interface at all? Because you cannot judge or steer work in a tool you cannot operate. When an agent returns a model that is the wrong size, floating, or twisted, you need to recognise it, and to say “scale it down,” “set the origin to the base,” “the normals are flipped.” Those are Blender’s verbs. Without them you are a passenger.

The good news: you do not need to learn to build a whole object by hand. You need a small vocabulary — how to look around the viewport and five core operations. That is enough to read what the agent did and name what to fix, which is the entire job of the human in this loop.

The next slide is a simulated Blender viewport. It is not the real program, but the moves are the real moves. Do them.

Do the five moves

Follow the guided steps below the viewport. Click each tool when it lights up; watch the object respond. Drag inside the viewport to orbit, exactly as you would orbit with the middle mouse button in Blender.

LayoutModelingSculpting(empty scene)
drag to orbit the viewport
Step 1 / 5Every Blender scene starts empty. Add an object — click Add Cube (in Blender: Shift + A).

This is a model of the interface, not the software — but “add, grab, rotate, scale, extrude, orbit” are the same actions you will use, and the same ones an agent performs for you when you prompt it later.

The five moves, and why each one matters later

Each operation you just performed reappears throughout the course — usually as something you ask the agent to do, then verify:

Add (Shift + A) creates objects — the agent does this on every “create a…” prompt. Grab (G), Rotate (R), and Scale (S) are the transforms; Scale in particular is the operation behind the wrong-scale failures in the very next module, and “apply scale” is the fix that shows up again at export. Extrude (E) builds new geometry from existing faces — how detail gets added. Orbiting is not an edit at all but it is your primary inspection tool: you cannot trust a single thumbnail, so you look from every side.

One more thing to recognise: the small orange dot on an object is its origin — the point it rotates and scales around and the handle used to place it. A wrong origin is invisible until you move the object, which is exactly why agents leave it in the wrong place.

Hands-on exercise

If you have Blender open, do the five moves there: Shift + A to add a cube, then G, R, S to transform it, Tab into edit mode and E to extrude a face, and orbit to inspect. If you do not have Blender yet, complete the guided sequence in the simulated viewport above and then free-play every tool.

Then write one sentence connecting a single Blender operation to evaluating agent output — for example, how knowing “scale” lets you catch the wrong-size model in the next module.

The failure gallery

Each of these is caught by a quality gate — keep the cheatsheet open while you work.

See the journey

🖼 Drop a screenshot or screen-recording of this case study here. screenshot slot · supplementary to the written core
The agent does the building; you supply judgment — but judgment requires vocabulary. Learn the five core operations and the viewport first, not to model by hand, but so you can read what the agent did and name what to change.

Cheatsheet

Prompt skeleton
Blender shortcut basics (keyboard): Shift + A = add object | G = grab/move | R = rotate | S = scale E = extrude | Tab = edit/object mode | Z = wireframe | numpad = views Middle-mouse drag = orbit | the orange dot = the object origin
Failure modes
  • Lost in the viewport, cannot find the object — frame it with Home or numpad-period
  • Operating in the wrong mode (object vs edit) — Tab toggles it
  • Confusing the origin (orange dot) with the object's centre
  • Transforming the wrong axis (click X/Y/Z to constrain)
  • Not knowing scale must be applied (ties back to wrong-scale agent output)
Key operations
  • Add an object (Shift + A)
  • Move / Grab (G), Rotate (R), Scale (S)
  • Extrude a face (E) in edit mode
  • Orbit / frame the view to inspect from all sides
Quality gates
  • Can you add, move, rotate, and scale an object without the menu?
  • Can you orbit to inspect an object from any angle?
  • Do you know what the origin (orange dot) is and why it matters?
  • Can you name the operation that fixes a wrong-scale model?
Workflow steps
  • Add an object to the empty scene
  • Grab / rotate / scale it to get a feel for the verbs
  • Extrude to see geometry being created
  • Orbit to inspect from every side
  • Map each move to "what the agent does for me later"
Next module
  • product_first_prompt — now you can read what the agent builds, start directing it.

Reflection card

Active retrieval — answer from memory before re-reading. Saved to this browser.

  • Completed the guided five operations in the interactive viewport (or in real Blender).
  • Can name each of G / R / S / E / Shift+A and what it does.
  • One sentence connecting a Blender operation to evaluating agent output.

Next: product_first_prompt — now you can read what the agent builds, start directing it.

Finish — back to Product Design →