You watched your student teacher run a full period. You have maybe ten minutes before your next class walks in, and they're standing there waiting for feedback. What comes out is usually "That went well!" or a vague note about pacing, because there isn't time to organize the three or four things you actually noticed into something useful.
Why the debrief usually falls flat
Good feedback for a novice teacher needs to be specific, prioritized, and actionable in their very next lesson. Under time pressure, you default to whichever comment comes to mind first, which is often the least important one. Meanwhile the pattern you actually noticed, like calling on the same four kids every time, or losing the room during transitions, goes unsaid because it feels harder to phrase kindly on the fly.
A faster way to organize your own notes
During the observation, jot fast, ugly notes: timestamps, quotes, what you saw. Right after class, before the next bell, paste those raw notes into a model and ask it to sort them into three buckets:
- What to keep doing — the one or two things that clearly worked
- What to try differently next time — the single highest-leverage change, not five
- A question to ask them, rather than a directive, so the debrief is a conversation and not a critique
This takes under two minutes and turns scattered notes into something you can actually say out loud in the hallway.
An example of the shift
Raw notes: "9:14 called on Marcus, Priya, Marcus again, Priya. 9:22 half the room off task during worksheet. Transition to groups took 4 min, lots of chair scraping."
Unstructured feedback: "It went well, maybe work on transitions."
Structured feedback: "Your explanation at the start was clear and the kids were with you. I noticed you called on the same two students most of the period; want to try popsicle sticks or a random-call app next time so more voices get in? And the transition to groups took about four minutes; what's one routine you could teach for that this week?"
The second version takes the same observation and turns it into something a student teacher can actually act on the next day.
Keeping the mentoring relationship, not outsourcing it
The model doesn't watch the lesson, doesn't know your student teacher's confidence level, and doesn't know if this is their third rough week in a row or their first bump. You're still the one deciding what to say and how to say it. What AI is doing here is the sorting and phrasing work under time pressure, not the judgment work.
If your program requires a formal training or onboarding document for student teachers on your team, our [training manual writer](/training-manual) builds a clear, skimmable reference from your program's requirements and your own routines, so the day-to-day norms your student teacher needs (how you want lesson plans submitted, how grading works, what to do if a kid is in crisis) are written down once instead of repeated verbally in bits and pieces across the semester.
What this actually buys you
Ten minutes between periods was never enough time to think from scratch. It's enough time to read three sorted bullet points and say them out loud with the right tone. That's the difference between a student teacher who leaves your room with something to try tomorrow and one who leaves with "keep it up."
