The concept
Each part answers one of four questions: what students make, why families care, where it leads, and what students decide.
Student outcome
Students train their own image, sound, or pose model with Google's Teachable Machine, a free browser tool with no accounts or code needed. Then they turn it into an invention: a recycling sorter that tells paper from plastic, a posture coach that notices slouching, a reader for sign-language letters, a pet-door gadget that recognises their dog. They leave with a working model that runs on a web page in their portfolio, on any laptop with a camera.
Why families would love it
- It feels like magic within minutes: the computer starts recognising what the student shows it. Children want to demonstrate it at home that evening.
- It works from age 8 with no accounts, which makes it easy to say yes to.
- It leads straight into science fairs, robotics clubs, and hardware projects, which families already value.
How it grows into a deeper project
- Train it: teach the model three things to tell apart, with examples the student collects.
- Break it: let friends try it and find where it fails, for example in different lighting, with different people, or with new objects. This leads into data and fairness: who did we forget to include?
- Improve it: collect better examples, retrain, and compare results before and after.
- Build the gadget: put the model in a web page, or connect it to a micro:bit so it lights up, beeps, or moves.
- Pitch it: present the invention to a mentor, or enter it in a science fair.
Students' own ideas and decisions
Students choose the problem their gadget solves, what it should recognise, and which examples to teach it. When it fails, they decide whether the fix is more data, different data, or a different design. Older students also decide when the gadget is "good enough" to use, and how to explain its limits honestly.
Open Teachable Machine →
How it adapts by age
Every age group invents something. What changes is who does the technical steps and how deep the thinking goes.
| Age | What students do |
|---|---|
| 8–10 | Train a model together in small groups (for example, happy and sad faces, or rock, paper, scissors) and invent a use for it. |
| 11–12 | Train their own model, test it with classmates, and improve it. |
| 13–18 | A full invention with a fairness test, before-and-after results, and an optional micro:bit build. |
Where the portfolio fits
Each gadget gets a project card on the student's own website, with what they built, who it's for, and what they learned. Over a term, the portfolio becomes evidence of real work for university and program applications.
See the featured Jarvis workshop, the full list of workshop ideas, and the portfolio session plan.