KDBM Lite · 04 / 10
Spec → Build → Check → Ship
Learn the four-step KDBM Lite loop that keeps an AI build small, checkable, and ready to share.
Spec → Build → Check → Ship
What this is
KDBM Lite uses one repeatable loop:
| Spec | → Build | → Check | → Ship |
|---|---|---|---|
| Define the result and boundaries. | Complete one small Work Card. | Compare the real output with the spec. | Save and share the strongest proof. |
The loop works for a content-led website and for a browser-local tool. The details differ, but the method stays the same.
Why it matters
A small spec gives both you and the AI a target. Without it, each prompt can pull the project in a new direction and make the next session harder to resume.
KDBM Lite stores the decisions in project files, not only in chat. That makes the work easier to continue, review, and hand over.
What to do
Use the loop at two levels:
- For the whole project: agree on a brief, architecture, design, and blueprint.
- For each change: build one Work Card, test it, update status, then choose the next card.
During this three-hour session the rhythm is:
| Time | Outcome |
|---|---|
| 0:00–0:18 | Understand AI building and how coding changed |
| 0:18–0:28 | Take the architect role and learn this loop |
| 0:28–0:43 | Learn the workspace and the AI helper |
| 0:43–0:55 | Download, open, and pass setup |
| 0:55–2:25 | Save a clear spec, then build Work Cards |
| 2:25–3:00 | Check, fix the smallest useful thing, and ship proof |
Common mistakes
- Writing a vague spec that cannot be tested.
- Building several cards at once.
- Shipping a preview that was never checked.
Next step
You now have the method. Before downloading anything, learn the cockpit you will build in: Your Builder Workspace.