How it works
A wrong answer is the most useful thing a learner can give us.
Most training marks a wrong answer and moves on. Ours stops, works out the idea behind it, and takes the learner down a short path built for that idea. Then it brings them back.
Arr! A black hole be a patch o' space where gravity grips so hard that not even light can sail away…
The idea
Branch on the misconception.
Every lesson has a main path. A wrong answer is never a dead end and never a red cross.
- The main path
- A path for one misconception
- The learner, wrong answer included

We name the mistake
Not "wrong", but "thinks a bigger bottom number makes a bigger fraction". Each lesson is built around the ideas learners actually hold.
The path is short and specific
A scene or two aimed at that one idea: a model that shows it failing, a question that exposes it.
Then they rejoin
The learner lands back on the main path with the idea fixed, not just the answer.
What learners touch
Models, not slides.
The learner doesn't read about the rule; they watch it work. Fraction bars to re-cut, a mini language model running in the page, a simulated web to fact-check against, a voice meter on their own microphone.
Arr! A black hole be a patch o' space where gravity grips so hard that not even light can sail away…
Echo
AI literacy you play, not read.
Tangent
Maths that clicks.
Your friend Wren saved for three weeks to buy the red kite. On its very first flight, the wind tore it.
Watch what Bram saysSkyGems Life Skills
Life skills as choose-your-own-adventure stories.
The lens
See how it's built, in every lesson.
Every training carries a "See how it's built" button. It opens the lesson's branch map, the principle behind each beat, the standards it targets and a live log of what the learner did. It is there for teachers, trainers and buyers, so you can check the thinking and not just the screens.



Memory across lessons
Later lessons remember.
Mistakes are not forgotten at the end of a lesson.
Echoes of earlier answers
In Echo, Episode 2 brings back the exact mistake this learner missed in Episode 1, then shows how it happens inside a real model.
A toolkit that grows
Prompts a learner saves in one episode come back when a later one needs them.
A finale made of their choices
The last Echo episode compiles the learner's own choices into personal AI rules and a certificate.
Narration
Every line is spoken.
A narrated guide reads the lesson sentence by sentence, with recurring characters and recorded voices where they're ready. Where a recorded clip doesn't exist, the browser's own voice takes over, so nothing goes silent.
Safety by design
Nothing collected, by default.
No accounts, no sign-up and no personal data. Progress stays in the learner's own browser. Read the safety page →
Our process
Six steps, from your topic to a finished training.
- 01
Map the misconceptions
We start with your learners: what they get wrong, and the idea behind it.
- 02
Script the branches
A main path, plus a short path for each misconception, each one bringing the learner back.
- 03
Build the models
Hands-on tools, so learners see the idea work before anyone explains it.
- 04
Test every path
Every branch is walked end to end, so no wrong answer leads anywhere broken.
- 05
Independent review
Someone who did not build the lesson checks it before it ships.
- 06
Narrate and launch
Recorded voices, your look, your own web address.
Want to see inside one?
Open any lesson and tap the lens.
Pick a course from our work and try lesson 1.
Bring us a topic.
We will map the misconceptions and show you a path.