The DEEP Build Kit · Card 01
The DEEP Build Loop
The method on one page: start with the change, not the tool; the five steps of the Loop and the block of the session for each; the complexity ladder; the three levels and the two door questions; the two rules.
For: Everyone
Card 01 · The method on one page · Hook · 15 minutes (0:00–0:15)
Start with the change, not the tool
Write down the change you want, and three success checks, before you choose a tool or build anything. Then judge the tool against your checks, not against how it looks. Everyone builds towards their own outcome, and is judged against their own success checks.
Teachers build with AI. Students open a finished page with no login, no personal data and no live AI.
"You can only prove that software failed to do what you asked if you wrote down, in advance, what you asked for."
Alex Gray, I Gave ChatGPT a Week to Build My App
The Loop: after the Hook, each block of the session belongs to one step
| Loop step | The question | Block of the session | Card | Move on when… |
|---|---|---|---|---|
| Brief it | What change do I want, for whom, and how will I know it worked? | 0:15 Define the outcome (25) · 0:40 Write the spec (15) | 02, 03 | Your canvas has an outcome, three success checks and your red lines, and your spec is written |
| Build it | What does version 1 look like? | 0:55 Build Sprint 1 (50), a thin slice, with the save point at 1:15 · 1:45 Break (10) | 04 | One success check passes, and that version is saved |
| Break it | Where is it wrong, unclear or unusable? | 1:55 Test like an engineer (20) | 05 | Your test partner has run your checks, and every failure is logged |
| Fix it | What is the one change that matters most? | 2:15 Build Sprint 2 (20): fix failed checks only | 06 | Each failed check passes, or is written into the spec for version 2 |
| Teach it | Who uses it next, and how will I know it worked for them? | 2:35 Gallery walk (20) · 2:55 Retro and commitment (5) | 15, 07 | Your commitment card has a date, two weeks from today, to test it with one real user |
It is a loop. What you learn in Break it and Teach it goes back into the spec for version 2.
The complexity ladder: what the tool does
Aim for one rung in version 1. Climb later.
| Rung | What the tool does | Example |
|---|---|---|
| Show | Presents something. The user reads or looks. | a labelled diagram, a worked example, instructions for a task |
| Respond | Reacts to what the user does. | a quiz with feedback, a simulation, a sorting activity |
| Remember | Keeps progress between uses, in this browser only. | progress through a set of questions |
A Remember build may store progress in this browser only, never what students type, with a Clear button. Test it where students will open it: browsers differ for a file opened from the computer, and a private window forgets it.
Think (AI inside the tool) is beyond this session: see card 18.
Three levels, sorted at the door
- Have you shared something you built with AI: a page or tool someone else opened?
- Have you used a coding agent, such as Claude Code or Codex?
| Level 1 ● Blue | Level 2 ▲ Orange | Level 3 ■ Black | |
|---|---|---|---|
| Your answers | No to 1 | Yes to 1, no to 2 | Yes to both |
| You start from | The build menu: a ready-made problem, outcome and checks (card 13) | Your own problem and the headed spec template | Your own problem, a blank page and two constraint cards |
| Tool | Canva, or any AI chat tool with a preview | Google AI Studio, or any AI chat tool | Claude Code or Codex |
| Your spec | Fill in the blanks | The headed template | A blank page |
| Your goal by the end of the session | A Show build, then a Respond build that passes its checks | One Respond build of your own, tested by someone else, shared as an .html file | A planned build, tested, then published or accessibility-checked |
| Stretch | Try the same spec in a second tool | Climb to the Remember rung, or publish a link | Publish it on GitHub Pages, or run the keyboard and contrast audit |
| Track card | 10 | 11 | 12 |
- Card 09 has the steps for each tool, including Claude, Gemini, ChatGPT and Microsoft Copilot, and a universal fallback (a .html file) for any AI chat tool.
- Level 3 ■ Black needs Claude Code or Codex working on your laptop (card 09 has the set-up). Not set up yet? Come to the install clinic at −0:15. If the install fails, work at Level 2 ▲ Orange (card 11).
- Switching is normal, up or down, at the save point or any time. Levels describe the task, not the person.
- Your mark is a colour, a shape and a number, on your badge and your track card.
The two rules
- No student names or data. Not in any prompt, file, file name, screenshot or tool. Use made-up examples.
- Nothing student-facing calls AI. Students use the finished page, never the AI tool. Every build today uses no AI and no keys.
Working alone?
- Pick your level by answering the two door questions yourself.
- Keep the order of the blocks. The times are a guide. Stop at the save point, and take the break.
- Set up your tool first, with card 09.
- Find a tester for Break it (card 05). Pair with anyone at a different level; if you can't, pair with someone who hasn't seen your build.
- Write your level on your canvas, in place of a badge.
DEEP Education Network · Build with AI · Two rules: no student names or data anywhere; nothing student-facing calls AI.