What a tutor does that a code generator does not
A generator answers the question you asked. A tutor works out the question you should have asked, then makes sure you can answer it next time without help.
| A code generator | A tutor |
|---|---|
| Produces code on request | Produces understanding you keep |
| Optimised for output speed | Optimised for what you can do afterwards |
| Explains when asked | Notices when you are lost and changes approach |
| Works on the text | Works on the running system, with evidence from your terminal |
| Finishes when the code compiles | Finishes when you can explain and defend it |
Both are useful. Only one of them changes what you are worth when the code breaks and there is nobody to ask.
What the LoWisa tutor can actually do
Press Start Tutor and it takes an active part in the project in front of you.
- Talks with you in real voice. Hands free, in the same session, so you can keep working while it explains.
- Opens your files and highlights the lines that matter. The explanation and the code stay on the same screen instead of in two different worlds.
- Draws the system on an interactive whiteboard. Architecture is spatial, and it is taught that way.
- Runs the terminal with you. Claims get checked with real commands, so you learn from evidence rather than assertion.
- Works in a safe sandbox. Experiments happen on a copy, so the project you care about is never the thing at risk.
- Checks that you understood. Every stretch of teaching ends with something you have to do or explain yourself.
A lesson, start to finish
This is the shape of an ordinary session on an existing project.
- It reads the project. A sandbox copy is scanned and a live map of dependencies and data flow is built, so nothing you are told is invented.
- It explains the architecture. Entry points, the path a request takes, and the decisions hiding in the code, drawn as it talks.
- It finds the gap. Not a generic curriculum: the specific thing in this project you cannot yet explain or defend.
- It sets the work. A hands-on task in the sandbox. Break it, watch what production would do, fix it yourself.
- It checks the claim. Commands run in your terminal decide whether the fix worked, and you read the output together.
- It tests ownership. You explain the system back. If the explanation has a hole, that is the next lesson.
Five ways to start
The tutor meets the project where it is. Whichever door you use, the destination is the same: you can explain it, change it and own it.
- System Immersion. A guided walkthrough of what the project does, how it is organised, its entry points and how the parts connect.
- The 25 ownership areas. Your project is scored against the areas that decide whether software survives, then the areas that will teach you most are sequenced first.
- Audit my AI code. Point it at a local folder or a GitHub URL and get the knowledge gaps, turned into a plan.
- Applied Mathematics Lab. Pattern recognition, probability, optimisation, systems thinking, trade-offs, decision making and root cause analysis, taught through real project situations.
- Project continuation. Come back to something half-built and pick the thread up without being re-taught what you already own.
Why the sandbox matters
Learning requires breaking things. Owning a project means you cannot afford to break it. The sandbox resolves that: the tutor works from a copy, so the honest version of a lesson, where you delete the index and watch the query time collapse, is available without consequence.
It also changes what the tutor is allowed to suggest. When an experiment is free, the teaching can be direct instead of cautious.
How it knows you understood
The tutor does not accept agreement as evidence. It asks you to explain the system in your own words, to predict what a change will do before making it, and to run the command that settles the question.
That is the whole product, in one sentence: the session is over when you could pass the interview about your own code.
The same tutor on your phone
The phone app is on Google Play and teaches the same way, lesson by lesson, with the tutor writing and speaking every lesson. There are structured paths for careers, programming languages and industrial systems, or any topic you name, plus learning cohorts where a creator plans the lessons, sets the day's work and runs contests, and Clock it, a shared feed where learners post what they discovered about AI.
A subscription follows the person, not the device: subscribe in the IDE and activate on the phone with the same email.
What it costs
Downloading is free and starts with 20 credits, awarded once on first install. LoWisa Starter is $8 a month with 200 credits included, LoWisa Plus is $11 a month with 500 credits included. A credit is one cent of AI, billed at exact provider cost.
Reading, drills and practice never spend credits. Tutor answers, lessons written for you and the studio voice do. Top up at any time: $5 adds 500 credits, $10 adds 1000 credits.
Frequently asked questions
Is this a chatbot inside an editor?
No. The tutor opens files, highlights lines, draws the system, runs terminal commands with you and sets hands-on work in a sandbox. Conversation is one of its tools, not the product.
Do I need coding experience to start?
No. It adapts from people who have never coded through to early-career engineers, teaching with real tasks and explanations kept close to the code.
Do I need my own AI provider key?
No. Both plans include the tutor with no key and no setup. If you would rather use your own provider account, BYOK is a $10 a month platform subscription and the link appears inside LoWisa when you connect your key.
Which platforms does it run on?
The IDE runs on Windows 10+, macOS 12+, Ubuntu 20.04+. The phone app is on Google Play.
Can I use it on a project I did not write?
Yes, and that is where it is strongest. Open an existing or inherited project and the tutor teaches it through System Immersion or the ownership areas, from a sandbox copy.