Ask
Who uses it, who may see what, and in which state must things stop changing? Read the request and acceptance criteria, then ask the AI in business language.
Tip: decide "who, what, and in which state" before you ask.
KASANE
A place to understand how it works, while AI builds it.
KASANE exists to help more people understand what happens inside—which technologies are used and how they work—while the built-in AI builds a business system for them. Once you understand how it works, you also know what to check in what AI builds.
THE NAME
The origin of the name "Kasane"
The name KASANE comes from "kasane-irome," the layered color combinations of Heian-era court dress.
It's an aesthetic of layering the outer and lining fabrics, and colors for each season, into a single outfit.
KASANE brings this idea of layering into how you learn. The system you have AI build is checked in three stacked layers: 1. behavior, 2. business pitfalls, and 3. key checkpoints. Layer by layer, you gradually come to understand what is really happening beneath "it works."
AI DEVELOPMENT PRACTICE
Ask. Check. Get it fixed.
You learn not how to write code, but—knowing what the AI is doing inside—what to tell it, what to check, and when to stop. You build that skill by actually having AI build, reading the explanations of its work, testing it, and getting results back.
Check
Try the finished system in the preview, and test it from another person's point of view too. Learn that "it works" and "it's safe to hand over" are different things.
Tip: hiding something on screen does not mean it can't be seen.
Get it fixed
For each failed item, a learning card explains why it matters, how to check it, and how to ask the AI to fix it. Have the AI fix it and resubmit as many times as you like.
Tip: describe what you want fixed in concrete business terms.
Inside the AI
See what the AI is doing, explained in plain words.
The AI's work is never a black box. The work log shows which part it is building right now—Screen, API, Database, Login & permissions. Every technical term that comes up gets a plain explanation with an everyday analogy, and if you want to know more, you can ask the AI right there.
Illustrative screen.
See what it's doing now
The work log shows which part the AI is working on—screen, API, database, login and permissions—so you can follow along without reading code.
Plain explanations for every term
Technologies that came up in your request—API, database, authorization—are listed and explained with everyday analogies.
Not sure? Ask the AI
Use "Ask the AI for more" to learn how a term is used in this very system, with the real files on screen.
Preview
What you ask for runs right in your browser.
Ask the built-in AI in plain words, and it builds the business system—ready to try in the preview next to it. Submit a claim, log in as someone else, and check it with your own hands before you submit.
Just ask in plain words
Tell the AI what to build and what to fix, in business language. No coding required.
Try it with your own hands
The finished system really runs, including saving data. Use the screens and test them from another person's point of view.
Safe to break
It runs in your own isolated environment. Rebuild as often as you like without touching any business data.
3-Layer Evaluation
Don't stop at "it works." Grading in three stacked layers
The name KASANE comes from the Japanese tradition of layering colors. We actually run your submitted system and check it in three stacked layers: behavior, business pitfalls, and key checkpoints.
1. BehaviorPublic
Does it work as the acceptance criteria say? You can read the acceptance tests in the problem
2. Business pitfallsHidden
Are business rules respected, such as approved claims not being editable?
3. Key checkpointsHidden
Security and governance: other people's data stays hidden, input fields can't be abused, password handling, approval records, and more
Passing all of layer 1 is required. The more of layers 2 and 3 you pass, the higher your overall rank
Let AI build it, and understand how it works.
Ask, watch it work, try it, and check. Experience the full loop with a practice problem.