Visualise the Flow.Understand the Learning.Explain the Intelligence.
EagLearn helps learners build AI projects while discovering how intelligence works: from data and models to predictions, experiments and real-world applications.
Every model is scored on examples it never saw Works with Arduino and ESP32 boards, or an on-screen robot
Learners do not just get an answer from a model. They build it, test it, and find out why it is right or wrong. Pick a stage to see it.
A model only knows what it is shown. Learners record their own examples, one class at a time.
An AI system is a chain of steps. Learners build it from blocks, and the tutor checks the chain is complete.
Training adjusts the model’s numbers so it makes fewer mistakes on the examples. It happens on the learner’s own device.
Part of every class is held back from training. The score comes from those examples, which the model has never seen.
The X-Ray shows what the model is given, what it turns that into, how it works on it, and why it decided.
After training, the tutor names the mistake in plain words, for example “it thinks your Triangle is a Star”, and suggests what to record next.
A star, read as a triangle.
More varied star photos, then train again.
Example numbers, to show the idea. Real results depend on the examples.
In Maker, the model’s answer can switch a light, turn a servo or drive a car on an Arduino or ESP32 board. With no board, an on-screen robot acts instead.
From recognizing patterns to building intelligent systems, every stage builds on what came before.
Patterns, sorting and colours, before any screen.
Sort picture cards into groups, then see a computer sort them the same way.
Knows that a computer learns from the examples it is shown.
Teach a computer to see, hear and recognise a pose.
Train an image, sound or pose model in one of three labs, with a tutor at every step.
Can say why the model got something wrong, and how more varied examples help.
AI with programming and robotics: 19 brains, 101 blocks, real boards.
Teach hand gestures, then use them to drive a servo, a light or a small car.
Understands the loop of see, think and act, and how a decision becomes a signal.
Inside the neural network: layers, neurons and how they learn.
The Picture Brain’s X-Ray shows its own filters and each layer’s maps.
Can describe what a layer does to a picture, from what it sees.
Language: sorting sentences, writing the next word, the meaning of words.
Word Brain, Story Machine and Word Galaxy, each with its X-Ray.
Knows that a language model picks likely words, which is not the same as true ones.
Explaining and questioning models: evidence, errors and limits.
X-Ray and Brain views, honest scores, and the table of what a model mixes up.
Can challenge a model’s answer with evidence from its own data.
Four projects that exist in the product today. Each one has a job for the AI, a job for the learner, and a test.
Every brain in Maker can be opened up. Learners see what it learned from, where it fails, and what its confidence really means.
Peek at Data shows exactly the examples it studied, so the learner can spot a missing or badly recorded class.
The results show what it mixes up, class by class, and tricks that fool it.
“80% sure” is how strongly the model leans, not a promise that it is right. A learner can set how sure it must be before a robot acts.
The tutor names the mistake and suggests what to record next. Learners retrain and compare.
The X-Ray shows when a model is guessing from the wrong clues, like the background instead of the object.
The same three answers as the screen above. “Happy” wins by 3 points: the model is close to guessing.
Being tested Every X-Ray with a live camera and microphone on ordinary school laptops.
Build an AI that knows your hand signs, then make it switch on a light or drive a robot.
Get StartedPhotos and sounds stay in the browser on your own device. Projects are saved as a file on your device. An account keeps a name and an email address.
How we handle dataA tutor in every lab says what to do next, and Maker has 66 chapters with questions and answers.
See a lesson, stage by stageRuns on the lab machines you already have. No installs, no admin rights, free during early access.
Bring EagLearn to your schoolBuilding a curriculum, a robotics kit or a learning platform? We would like to hear from you.
Talk to usThese are three of the 22 starter projects that come with Maker. They are not learner work: when schools share learner projects with permission, they will appear here.
If the machines in your lab open a website, they run EagLearn. Here is what a school needs to know.
EagLearn is in early access. We show quotes only from real, named teachers and parents who agree to it. Until then, here is how to check what we say.
Models are trained in the browser, not on our servers. Maker keeps working offline after the first visit.
The test examples come from the end of each recording, so near-identical frames cannot leak from learning into testing.
Photos and sounds stay in the browser. Close the tab and the recordings are gone.
Each X-Ray is drawn from the running model, and automatic tests check its arithmetic against the real one.
Anything built but not yet proven on real hardware is marked “Being tested”, here and in the app.
About data, safety or results. We answer in writing at eaglearn.info@gmail.com.
Start with one journey, or bring EagLearn to a whole school. Free during early access.
