Meet the tutor
A teacher that knows your child, remembers them.
Built on the same generative AI you've heard of — constrained, supervised, and shaped into something a child can actually learn from. Six concrete behaviors, verifiable in the first week.
Six behaviors that distinguish a tutor from a chatbot
What you'll see in week one.
None of these are aspirational. Every one is a behavior you can test the first time your child opens the academy.
01
It knows your child by name and by frontier.
The tutor opens every session with what your child did yesterday and what's planned today. It calls back to specific things they said last week. It remembers that they love horses and that subtraction is currently hard. This is not metaphor — it's how the system is built.
02
It asks before it explains.
Ask a math question and a chatbot answers. Ask a math question and a tutor asks what you've already tried. The L1fe tutor is taught to surface your child's thinking before delivering an explanation — because the explanation is wasted if the question hasn't been articulated.
03
It will not move on prematurely.
If your child types 'I get it' after thirty seconds on a hard concept, the tutor checks. If the check shows they don't, the tutor stays. Class periods are an artifact of a one-teacher-to-thirty-children physics. We don't have that physics anymore.
04
It lives inside a private dataspace per child.
Every conversation your child has with the tutor is recorded into a per-child memory store that only your family can read. The tutor cannot share information across families. It cannot use your child's data to train a model that talks to other children. This is enforced at the system level, not as a policy.
05
It cannot browse the open internet.
The tutor has access to a curated knowledge base — vetted reference texts, age-appropriate readings, the academy's own curriculum. It cannot pull from the open web in the middle of a lesson. There is no path for your child to be sent to a Reddit thread.
06
It refers value-laden questions to you.
Ask the tutor about religion, politics, sex, or anything else where reasonable parents disagree, and it will say so — clearly, by name — and offer to surface multiple perspectives or to wait until you can talk together. It will not pretend to be neutral when it isn't.
It waits
Not a chat-bot. A patient teacher.
The tutor sits with the question. Lets the silence work. Asks before it explains.
What we built it on
No mystery. No black box. The shortest honest explanation.
The tutor uses the latest generation of large language models (the same family of technology behind the AI you may have tried at work) — currently Anthropic’s Claude, run under a Zero-Data-Retention contract that prohibits use of your child’s conversations to train any commercial model.
On top of the language model, we built a teaching system: a per-child memory store, a curriculum scaffold, a daily lesson generator, a posture controller that decides when the tutor should ask vs. explain, a safety layer that handles value-laden topics, and a parent-facing audit trail.
We are explicit about what is and isn’t the language model. The model is the engine. The teaching is the rest of the system, designed by people, around your child.
Hear it for yourself
A sample conversation, on the application call.
Real prompts, real responses, with your child's grade and interests in mind. Twenty minutes, no slides.