Coaching the adult who teaches, rather than tutoring the child

Context

My wife home-educates our children, and for two years she had been logging each day's lessons in a paper planner. It held almost everything she knew about how the teaching was going, and nothing ever read it back to her. Paideia started there: an AI coaching tool for home educators, designed and built solo in 2026, with that household as its only pilot.

The problem

Home educating is not a smaller version of school. There is no department to compare notes with, no scheme of work to measure against, and no external mark of whether a year went well. The tools that exist mostly assume otherwise.

• Working alone. No colleague to think out loud with, and in a home-educating context admitting that a lesson went badly carries a weight it would not carry in a staffroom.
• A record that only records. The planner captured each day faithfully and gave nothing back, so reflection depended on memory and on whatever she happened to notice at the time.
• No view of the arc. Nothing showed how a child had moved across a term or a year, which matters more as the legislation around home education tightens.

Research and insight

The tools split cleanly into two bands. Planners point forward at the week you intend to teach, and direct-to-child tutors replace the adult altogether. Neither does anything for the person doing the teaching.

What convinced me the space between them was real rather than merely empty was Erik Hoel's essay on why we stopped making Einsteins, which argues that one-to-one tutoring is the only method that has reliably produced them, and that it vanished because it cannot be standardised. Einstein had Max Talmud. The practice is old and reasonably well evidenced. It had simply never been productised.

Strategic reframe

She was already logging retrospectively, and using that logging as her reflection. Designing forward would have meant competing with planners and asserting what a week ought to contain, which in home education is the wrong posture, because the approach is tailored to the child by definition. So the tool reads what actually happened and coaches from that, which makes it empirical rather than prescriptive.

Design decisions

Most of the decisions that mattered were about restraint rather than capability. The model can do considerably more than the product lets it do, and almost every rule below exists to stop it doing something it would otherwise do fluently and wrongly.

A teaching assistant, not a teacher

A tool that tells a home educator what she ought to be teaching is a red rag, because not being told what to do is often part of why a family is home educating in the first place. So Paideia sits as a teaching assistant rather than a teacher, a coach or a mentor. It can notice things and ask about them, and it can suggest a way to consolidate something she has taught, but it never decides what comes next.

The burden is not the seconds, it is the doubt

Documentation tools are usually judged on effort, so the obvious target is fewer taps. The pilot made it clear that the felt cost was not the seconds, it was not knowing whether any of it had been worth doing. So the rule became a ratio rather than a race: take the input from artefacts that already exist, starting with a photograph of the paper planner, and spend the output on showing her what she structurally cannot see.

A conversation of length one is complete

Capture became a conversation rather than a form, because a form receives what it is given and a conversation can notice what is missing. The risk is that a surface which can ask drifts into one that expects an answer, so the floor is explicit: she can open it, drop in a photograph, say nothing and leave, and that is a complete record. I track the proportion of one-line conversations, because if it ever reaches zero the floor has stopped being real.

A bare loading spinner is a bug

Three eight-second waits in one sitting are not remembered as twenty-four seconds, they are remembered as a tool that always makes you wait. So every wait is either removed from her path, hidden behind something she was going to do anyway, or filled with an honest account of the work actually happening. That last option is the most tempting and the most dangerous, because the first time an animation is caught overstating what it is doing, every honest one after it reads as theatre.

The record has no prose field

An early version praised a child's progress on a fortnight of thin data, and it read as flattery, which cost more than saying nothing would have. The fix was structural rather than a change of tone. The generated record has no free-text field at all: it is assembled from coverage, typed observations and quotes copied out by code, and every entry carries a reference back to the thing it came from. Where her account and the worksheet disagree, both go in, unresolved.

Building it

I built it solo. Expo and React Native on the front, Supabase for data and auth, the Claude API for extraction and synthesis, deployed as a progressive web app so it installs on a phone without an app store. The hard part was never the interface, it was the schema. A worksheet page is not one piece of evidence, it is twelve, and any claim the tool makes months later has to be able to point back at the right one.

Constraints and trade-offs

Two things I decided not to build. The first is curriculum advice. Telling a tutor what a child ought to learn next makes the tool the authority, and the premise of the whole thing is that the tutor is. An educational philosophy could arrive later as a lens someone opts into, but never on the main path.

The second is anything touching additional needs. I talked it through with a child educational psychologist I know and concluded that the distance between noticing a pattern and implying a diagnosis is shorter than it looks, and not a line for a tool at this stage to walk up to.

The pilot is one household, so the findings are deep rather than broad. There is also a limit I recorded during the build and have not solved: a system fed only one person's account can enrich that account, but it can never surprise it.

Outcomes

There are no commercial outcomes here. There is one household, four moderated sessions across ten weeks, and a set of accept criteria written down before each one so that a disappointing result would be a finding rather than a matter of interpretation.

The most useful thing the pilot produced was a no. At the end of the third session I asked whether the logging had been worth doing, and the first answer below is what came back. The second is the same person on the same product five weeks later, after the thing she had actually been describing got built.

Reflections

Managing designers, I could always describe a constraint and hand it over. Building this alone meant every decision I made about posture had to survive contact with a schema, and a few of them did not. The rule that the record can only say what the evidence supports sounds like a values statement right up until you are the one writing the table that enforces it, at which point it becomes a column and a migration.

The other thing I had not expected to spend so much time on was the law. Holding information about a named child is legally fraught, and working through it turned what had been a design preference into the thing that makes the product defensible: Paideia does not assess the child, it supports the adult teaching them. It records what happened in a lesson, but every output is aimed at the tutor's practice rather than at a verdict about her child. Parents guard information about their children closely, and that instinct turned out to be a sharper design constraint than anything I would have reached on my own.

What is unresolved is whether any of it generalises. One household produced findings I trust and a sample I do not, and the only way to know which parts are about this family and which are about home education is to put it in front of households I am not part of.

The practices behind this →