PiPGAME
STUDIO

THE MAKING-OF / A CLASSROOM FIELD GUIDE

An idea.
A little AI.
A lot of “try again.”

How Anderson and family are turning game ideas
into PiP Studio—and what we learned along the way.

Let’s open the notebook

A story about engineering. No coding experience needed.

PIP’S WORKSHOP NOTEBOOK
KEEP THE QUESTIONS COMING.
THE BIG IDEA

AI can help us build. We still have to choose, check, and improve.

01

START WITH AN IDEA

“What if we made a game?”

Anderson loves making games. PiP Studio is a place for those ideas: different worlds to play, a fluffy axolotl guide, and a workshop for dreaming up new characters.

For a class presentation with his mom, the plan grew into a collection of three games. But a big idea is easier to build when we ask for one small step at a time.

02

PEOPLE SET THE DIRECTION

The AI guessed.
Anderson corrected it.

At first, the AI suggested a gentle reef-rescue game for Gills and Glory. That was a proposal. It was not the game Anderson wanted.

FIRST AI DRAFT

Explore and rescue.

Swim around a reef and help little fish.

ANDERSON’S DIRECTION

A playful fighting arena.

Battle a computer opponent, or play against another person.

The design changed. A clear correction can matter more than a longer prompt. People decide the goal; AI helps work toward it.

03

GIVE THE IDEA A SHAPE

PiP needed more than a picture.

A character in a 3D game needs a shape that can be seen from different sides. Blender helped create original 3D props, including PiP, coral, rocks, and a shell arch.

Unreal imported those props into a real game scene. The browser version can use the same art in a different file format. The tools have different jobs, but they can share the same idea.

IdeaWhat should it be?
BlenderMake the 3D shape.
GameMake it move and play.

The art source is saved, so someone can open it and make another change.

04

ASK FOR EVIDENCE

“It ran” is not the same
as “it worked.”

One Unreal command ended without reporting a problem. But the file it was supposed to create was missing. We could not call that a successful build.

The protected workspace had stopped the script from doing its job. After the run was allowed in the right place, the engine produced the saved scene and a build record. We checked the result, not just the message.

A BETTER QUESTION

What should exist when this works—and can we show it?

A browser also reported a missing little tab icon. Small errors count too: add the missing file, then check again.

05

TEST LIKE A PLAYER

The button was there.
The click couldn’t reach it.

In the copied Spy Kids game, a mission-preview heading covered part of the launch button on a desktop screen. The button looked ready, but another part of the page caught the click.

Keyboard launch worked. That still did not prove mouse launch worked. The heading was moved, then an ordinary click was tried again on a desktop-size screen and a phone-size screen. Both checks reached the opening dialogue.

One successful path does not prove every path works. That short check also does not prove the whole game has been completed.

06

MAKE A SMALL LOOP

Change. Check. Learn. Repeat.

An iteration is one small round of improving something. Our plan allows up to ten rounds. A round needs its checks before the next round can count.

  1. 1PlanPick one small change.
  2. 2BuildMake that change.
  3. 3PlayTry the real controls.
  4. 4CheckCompare with the goal.
  5. 5FixRepair what failed.
  6. 6RecordSave what we learned.

A fast computer player can help check many moves. It cannot tell us whether the game feels fun to our classmates. That still needs people.

07

NOW IT’S YOUR TURN

Can one word change an idea?

Build a character prompt below. Read it out loud. Then change just one choice. What do you predict will be different?

YOUR PROMPT

A friendly teal robot with bubble shoes, in a colorful clay game style.

This is a writing experiment. It does not send a prompt to a model or make an image.

In the studio workshop, an adult can help you compare actual character results. Use original ideas and a made-up alias. Keep private details out of prompts.

A TEAM WITH DIFFERENT JOBS

Who did what?

PEOPLE

Choose the world.

  • Bring the game idea and PiP’s identity.
  • Correct the AI when it guesses wrong.
  • Decide what feels fun, kind, and clear.
  • Try the games on real classroom devices.
AI ASSISTANCE

Help make and check.

  • Draft code, art scripts, and explanations.
  • Turn a small request into files.
  • Run automated checks and record results.
  • Suggest fixes for people to review.

AI can make mistakes. A confident answer still needs a check.

FIVE WORDS TO TAKE WITH YOU

The little engineering dictionary.

Prompt
Words or choices that ask an AI to do something.
Model
A trained computer system that uses patterns to make an answer.
3D prop
A digital object with a shape you can look at from different sides.
Test
A check with an expected result. Did the thing we wanted actually happen?
Iteration
A small change, a new check, and a note about what we learned.

FOR THE PRESENTATION / ABOUT TWO MINUTES

A short talk track.

  1. 0:00–0:20

    Start with the idea. Anderson is making a little game studio. Show PiP and name the three games in the collection.

  2. 0:20–0:45

    Show that people steer. Explain the rescue-game guess and the correction to a playful fighting arena. AI’s first answer is a draft.

  3. 0:45–1:10

    Show one real mistake. A button looked ready, but another panel caught the click. We changed the layout and tried the actual click again.

  4. 1:10–1:35

    Try one change. Ask the class to predict what changing only a character’s color will do. Read the two prompt versions.

  5. 1:35–2:00

    End with the loop. Plan, build, play, check, fix, record. Explain what is checked so far and what still needs testing, then open a game.

Before class: test the exact school device, browser, projector, network, controls, and sound level. Use short demo routes. The recorded browser checks do not replace that rehearsal.

Open the collection