Every tool in this category is adding a button that writes the parent update for you. We are not going to, and since that will look like a missing feature, it is worth saying why.
The message is the product
A tutor’s parent update is a claim about a specific child, made by a named professional, to someone paying for an outcome. Its whole value is that a person who was in the room is telling you what they saw.
A generated version of that message is a different object wearing the same clothes. It reads better than most tutors write, which is exactly the problem: it is fluent about a lesson it did not attend, and fluency is what parents use as evidence of attention. The better it writes, the more confidently it says things nobody checked.
And the failure mode is not an obvious one. It is a sentence that is plausible, warm, specific-sounding and slightly untrue — “she’s grown in confidence with quadratics” when the session was about simultaneous equations. A parent cannot catch that. The tutor skims it, recognises their own general sentiment, and approves. The error enters the record and gets repeated, because the next summary is generated from the last one.
Where we do use it
The line is not AI versus no AI. It is who the output is for.
AI drafts things for the tutor, who is in a position to know whether they are right:
- a session log dictated on the walk home, turned into structured fields
- a suggested next step on a goal ladder, based on the last few sessions
- a weekly focus idea
Every one of those lands in front of the tutor as a draft they edit and approve, and nothing is saved until they do. The tutor can tell whether “next step: two-step equations with negatives” is sensible, because they taught the lesson.
What a parent receives is assembled from fields the tutor actually filled in — deterministic templates, the same inputs producing the same output every time. There is no model in that path. If the summary says a student worked on simultaneous equations, it is because a human typed that.
The side effects
Two things fall out of this that are easy to miss.
Private notes cannot leak, because there is nothing that could paraphrase them into an outbound message. A tutor’s own notes — “mum seems stressed about the fees”, “he mentioned the divorce again” — never go into the summary path at all. That is enforced in the code and covered by tests, not left to a prompt.
The record stays worth something. Six months of summaries is a real longitudinal account of a child precisely because each entry was written by someone who was there. Generate them and you have six months of plausible prose about a student, which is worth nothing when a parent asks what changed since August.
The honest cost
This makes TutorWrap more work than the alternative. You fill in a short structured form after each session; a competitor offers a button. We think the form takes about two minutes and is the only reason the output means anything, but it is a real cost and we are not going to pretend otherwise.
If what you want is something that writes convincing parent messages without your involvement, that product exists. It is just not this one.