What actually happens in a session.
Most learning apps are a bank of questions with a progress bar on top. Abbi is built as a loop — choose, teach, adapt, record, report — so that each session starts from what really happened in the last one.
A session, start to finish
Sessions are short on purpose — around ten minutes of focused work is worth more than an hour of drifting. Here is the shape of one.
- Abbi picks one thing
Not a menu. Abbi chooses the next objective from your child’s actual record and what it builds on, so a child never opens the app and has to decide what to do.
- It is taught, not just tested
The concept is explained inside a teaching pattern chosen for the objective, then practised. A child works through a sequence, not a single demo question.
- Help arrives, then steps back
When your child gets stuck, support increases — a hint, a worked first step, a visual cue. As they get it, the support is withdrawn, because doing it unaided is the point.
- Abbi watches how, not just whether
How long a step took, how many attempts, whether a hint was needed, whether the answer changed. Signals are combined, never read one at a time as proof.
- Evidence is recorded, not assumed
Getting one right is not mastery. Progress moves when there is enough independent evidence, which is why a good week and a lucky guess do not look the same in the record.
- You get told what happened
A plain-language summary after the session: what was practised, where they got stuck, what improved, and what Abbi plans to do next.
It remembers, so your child does not start over.
The difference between a tutor and a worksheet is continuity. A tutor walks in already knowing that last week the tens column was the sticking point, that your child gives up faster when the question is wordy, and that showing the first step once is usually enough.
Abbi keeps that picture across sessions and teaches from it. So the next session opens where the last one actually left off — not at the top of a syllabus.
Crucially, the record is not a profile the model writes about your child on a whim. What gets remembered goes through the system, not the conversation. How we use AI covers exactly where that line sits.
-
Level and language are separate
A bright six-year-old working two years ahead gets harder material without being talked to like an older child. An older learner filling a gap gets the right content without being talked down to.
-
Practice is not performance
Abbi distinguishes work done with help from work done alone. Only the second kind moves a skill forward.
-
Stopping is allowed
A child can pause and come back without losing anything or being nagged. Nothing in the design punishes a break.
-
Uncertainty stays uncertain
A pause is not evidence of confusion. Abbi holds the unknown as unknown rather than turning it into a confident claim about your child.
What you see afterwards
After each session you get a summary in plain English — what your child practised, how much help they needed, where they got stuck, and what comes next. Weekly and monthly views compare like with like, so "needed two hints this week against four last week" only appears when the work was genuinely comparable.
What you will not get is false precision. If Abbi is not sure, the report says so. See what the parent view looks like →