← All posts
April 3, 2026·10 min read

AI notes for teachers: parent evenings, lesson observations, and safeguarding records

Parent evening is 24 five-minute appointments written up from memory. Lesson observations lose specificity between the classroom and the debrief. Safeguarding disclosures require exact contemporaneous records. Here's how on-device AI changes the documentation problem for teachers.

Key takeaways
  • Parent evening: 20+ five-minute appointments documented from memory. By appointment 15, the earlier conversations are available only in outline. AI capture produces accurate records for all of them before you leave the building.
  • Safeguarding disclosures require the child's exact words recorded contemporaneously. A voice memo immediately after the disclosure is a stronger evidentiary record than a paraphrase written hours later.
  • EHCP review meetings with multiple professionals are complex and generate attributed content the SENCO cannot capture while chairing. Speaker diarization separates each professional's contribution.
  • Most school buildings have poor mobile data penetration. Kuulo works entirely on-device — reliable in classrooms, corridors, and meeting rooms regardless of connectivity.

Parent evening is three hours of five-minute appointments, back to back, twenty-four sets of parents, each wanting to know how their child is doing, what they need to work on, and whether you've noticed the thing they've been worried about at home. You have the data — you know these children, you've marked their books, you've watched them in lessons for months. The conversation is the easy part.

The hard part comes after: writing up twenty-four sets of notes before the details collapse into each other. By appointment fifteen, you are already reconstructing the earlier conversations from impression rather than memory. By the time you get home, the specific thing you said to the Sharma family about Mohammed's reading and the specific thing that Mrs Kowalski mentioned about Anya's friendship group are available to you only in outline.

AI capture changes this. Not by replacing the conversation — the conversation is the point — but by making the documentation as accurate as the conversation itself.

The teacher documentation problem

Teachers generate a continuous stream of notes that matter and are difficult to produce accurately. Parent meeting records. Lesson observations. Staff meetings with action points. SEND review meetings with multi-agency attendees. Safeguarding referral records where the child's exact words have legal significance. CPD sessions where the content is relevant to practice.

All of these share the same structure: a live encounter that generates content, followed by documentation that tries to reconstruct the content from memory. The gap between the encounter and the documentation is the gap between what happened and what the record says happened.

For most of these documentation types, accuracy is not merely good practice — it is a professional or legal requirement.

Parent evenings: the specific problem

A parent evening note serves multiple purposes. It is a record of what was communicated to the parent. It may be referred to if a parent later disputes what they were told. It is the basis for any follow-up action — an EHCP referral, a pastoral concern flagged for the head of year, a reading intervention agreed with the parent. And it is the record that a subject leader or form tutor draws on when making decisions about the student.

The standard approach — write brief notes during the appointment, expand them later — fails at the volume and pace of parent evening. Brief notes become cryptic. "Struggling — check in with SENCO" without context becomes ambiguous three weeks later. "Parents concerned about friendship group" without detail is unactionable.

A Kuulo workflow: keep the phone on the desk, record each appointment. After the last parent leaves, generate the summaries. Twenty-four appointment records, each covering what was discussed, what concerns were raised, and what was agreed, produced before you leave the building.

The appointments are on the record not as you reconstructed them but as they happened.

Lesson observations

Lesson observations — whether by a line manager, an external inspector, a peer observer, or the teacher observing their own lesson — generate feedback that is time-sensitive and context-dependent. The observer who tries to write comprehensive notes during the lesson is distracted from observing. The observer who writes up from memory after the lesson loses the specific language of the moments that mattered.

"Around fifteen minutes in, when the teacher moved to the second task, three students at the back lost engagement immediately" is a more useful observation note than "engagement dropped in the second task." The specificity comes from having captured the moment.

For Ofsted-style observations, the specific language of what was said, what was asked, how students responded, is the difference between useful developmental feedback and generic commentary. A teacher receiving specific observation notes can engage with specific feedback. A teacher receiving reconstructed impressions is working with a weaker quality of evidence about their own practice.

SEND and multi-agency meetings

Education, Health and Care Plan (EHCP) review meetings are complex, multi-attendee, and high-stakes. They may include the class teacher, the SENCO, the parent, the child (depending on age), a speech and language therapist, an educational psychologist, a social worker, and other professionals depending on the child's needs.

Every participant has a contribution to make. The SENCO chairing the meeting cannot simultaneously chair, engage with parents, and document comprehensively. The resulting minutes are typically written from the SENCO's notes, which reflect what they captured while managing the meeting — not a complete record of what was said.

Speaker diarization in a multi-agency EHCP meeting separates the contributions of different professionals. The SALT's targets, the EP's recommendations, the parent's account of progress at home — all attributed and distinct in the record. The EHCP document that emerges from a meeting with an accurate attributed record is better informed than one drawn from incomplete notes.

Safeguarding referrals: the exact words matter

When a child discloses something to a teacher — and disclosures most often happen informally, in a corridor, at the end of a lesson, in a quiet moment — the teacher's first responsibility is to listen without questioning. The second is to record what was said as soon as possible, using the child's exact words.

"In the child's own words" is not merely best practice guidance. In proceedings that may follow — police investigation, social care assessment, family court — the original disclosure record is evidence. A paraphrase written two hours later, under time pressure, from memory, is a materially weaker form of that evidence than a contemporaneous note made immediately after the conversation.

A voice memo into Kuulo immediately after a safeguarding disclosure — "Year 7 pupil, Amara, said to me after the lesson: [quotes what the child said] — generates a timestamped, structured disclosure record. The timestamp establishes contemporaneity. The exact language preserves the child's account as given.

No evidence from a web search or training material is being put forward here as legal advice. But the principle — that exact contemporaneous records are preferable to reconstructed paraphrases — is established practice guidance from the Department for Education's Keeping Children Safe in Education and is consistent with what teachers are told in every safeguarding training they attend.

Staff meetings and action items

Staff meetings produce action items. Someone agrees to do something by a certain date. The action item is written on a whiteboard, captured on a slide, or noted by whoever is taking the minutes. In practice, the minutes are often incomplete, circulated late, and not checked against what was actually agreed.

A teacher who records a staff meeting and generates an attributed summary knows what was agreed and by whom. When the HoD checks in on whether the intervention programme has been set up, the teacher who recorded the meeting can confirm both the commitment and the rationale discussed.

GDPR and pupil data

School data is subject to UK GDPR. Pupil data that relates to health conditions, special educational needs, safeguarding concerns, or family circumstances is special category data — Article 9 — requiring heightened protection.

Most AI note-taking tools require audio to be transmitted to cloud servers for processing. A recording of an EHCP review meeting or a safeguarding conversation, sent to a cloud server for transcription, creates a data trail for special category pupil data that the school's DPIA may not cover.

On-device processing means the audio stays on the device. The GDPR position is substantially simpler: the data is processed on the school device in the teacher's possession, consistent with the school's existing data handling policies.

This is the same reason on-device processing matters in clinical contexts, legal contexts, and research contexts: it eliminates the cloud processor from the data flow, reducing both the legal exposure and the practical risk of breach.

The offline requirement

Many school buildings have poor mobile data penetration. Wifi exists but may be restricted to staff areas. Classroom connectivity is often unreliable. Cloud-based transcription tools that require internet — Otter, Fireflies, Granola — fail in exactly the rooms where the recording is most useful.

Kuulo runs on the Neural Engine of the iPhone, regardless of connectivity. A lesson observation in a mobile-data-dead classroom works the same as one in the staffroom with full wifi. The app does not buffer for later upload — it processes immediately, on the device, producing the note before the observer leaves the room.

What this changes in practice

The teacher who records parent evening, lesson observations, and multi-agency meetings is not doing more administration. They are doing the same administration more accurately, in less time, from a better quality of source material.

The parent who raises a concern at parents' evening and follows up two months later will find the record of what was discussed is accurate and specific. The SENCO who refers back to the EHCP review minutes will find a complete attributed account. The line manager reviewing the lesson observation notes will find specific, detailed feedback rather than reconstruction.

The documentation was always required. What changes is whether it reflects what actually happened.

Frequently asked questions

Can teachers use AI to take notes at parent evenings?

Yes. Kuulo records each parent appointment on-device, with no audio leaving the device. After the last appointment, the AI summary covers every conversation — what was discussed, concerns raised, and actions agreed — produced before you leave the building.

Is recording a parent meeting GDPR compliant in a UK school?

UK GDPR applies to parent meeting records containing pupil personal data. On-device processing means audio stays on the teacher's device — no cloud processor receives it, which simplifies the data governance position significantly compared to cloud transcription tools. Parent notification of recording remains the school's responsibility under its privacy notices.

How should teachers record safeguarding disclosures?

Best practice (Keeping Children Safe in Education) requires the child's exact words recorded as contemporaneously as possible. A voice memo made immediately after the disclosure, generating a timestamped AI-structured record, is a stronger contemporaneous record than notes written hours later from memory.

What app works for lesson observation notes?

Kuulo records the observed lesson on-device, producing detailed notes including specific moments, student responses, and timing — more specific than notes written from memory after the observation. Offline operation means it works in classrooms without wifi or mobile data.

Try Kuulo

On-device AI notes, private by design. Free for iPhone and Mac.

Get the app