What happens when schools cannot give every teacher a coaching opportunity?
I am a big fan of instructional coaching. When done well, a skilled coach working alongside a teacher agrees on a precise focus, gathers evidence and discusses classroom practice to help the teacher make small, deliberate improvements over time.
The keyword is skilled.
Instructional coaching requires time, trust, expertise, confidentiality and a shared commitment to improving teaching. I have written before about Jim Knight’s work on instructional coaching, where partnership places the teacher firmly at the centre. I’ve also shared many resources to help schools set up coaching successfully.
Why Does Coaching Stall?
Schools often start coaching programmes with ambition. Then unforeseen logistics stall. Coaches are stretched. Timetables clash. Cover is tight. Priorities shift. Staff don’t commit. What started weekly becomes fortnightly, monthly or occasionally.
The problem is not necessarily the coaching model. It is the infrastructure required to sustain it across an organisation. Research from Harvard’s Center for Education Policy Research highlights the same challenge: coaching is labour-intensive, with limited human resources restricting how regularly teachers can receive feedback.
Instructional coaching is too important to be rationed to a few teachers, a few times a year.
Can AI Coach Teachers?
When people suggest that AI can “coach teachers”, my first response is scepticism.
- Can AI understand the context of a department?
- Can it read the relationship between teacher and pupil?
- Can it understand the emotional temperature of a classroom?
No. Not in the way a great human coach can.
The stronger argument is not that AI replaces instructional coaching, but that technology might help schools extend it. That is the problem Starlight is trying to solve, and why I’m happy to advocate the platform.
Coaching Between Coaching
Starlight is a UK-built AI coaching platform. A teacher records a lesson, uploads the audio and receives a structured whole-lesson report within minutes. The human coach still matters most.
A coach and teacher can agree the development focus together. Between meetings, the teacher uses Starlight to gather lesson evidence. They can then return to the next human coaching conversation with something more precise than memory: a transcript and report to discuss. This is not replacing the coach. It is giving the coach better, raw material.
Trust Comes First
If Starlight became a surveillance tool, I would not support it. Instructional coaching cannot thrive where teachers feel monitored. It depends on psychological safety, professional agency and confidentiality.
Starlight’s model is designed around that principle. Reports remain private to the teacher. School leaders see anonymised, aggregated trends only. Individual teachers are not named in the leadership view. Teachers can delete their data, and the school remains the Data Controller–that distinction is critical.
- For leaders, the opportunity is more frequent reflections without adding more observations to the calendar.
- For coaches, AI-generated lesson evidence could make the human conversation sharper.
- For teachers, it offers regular feedback between coaching sessions without waiting weeks for the next meeting.
Can Instructional Coaching Scale?
This is where AI may have a legitimate place in teacher CPD: not as a substitute for professional expertise, but as a tool that helps instructional coaching reach further.
AI cannot replace the trust, expertise and professional judgement of a skilled human coach. But it may help teachers and coaches see more, remember more and discuss classroom practice more precisely.
Instructional coaching works. The next challenge is helping it scale without losing what makes it powerful in the first place: trust, expertise and the human conversation.
Coaching stays human. AI provides the evidence.
TeacherToolkit readers can try Starlight free this term.
This is a promotion.