The 4pm Exit-Ticket Review That Runs Itself
Scheduled tasks in Claude for Teachers let you hand off reviewing each day's exit tickets and adapting tomorrow's plan — running every school day at 4pm, so the analysis is done by the time you get home.
The exit ticket is one of the best formative-assessment tools ever invented and one of the most wasted. The idea is airtight: at the end of class, students answer one quick question, and their answers tell you whether the lesson landed. The failure is entirely in the follow-through. To be useful, the stack has to be read tonight — while it can still change tomorrow's lesson. And tonight you have thirty other tickets, a family, a commute, and a plan for tomorrow that's already half-written. So the stack sits, and by the time you get to it the moment has passed.
Claude for Teachers offers a genuinely different answer to this: don't read the stack faster, hand off reading it entirely. Because the product includes Claude Code and Cowork, Claude can carry multi-step work forward on its own — and it can run on a schedule. The launch names this exact use case: a scheduled task that reviews each day's exit tickets and adapts tomorrow's plan, running every school day at 4pm. Claude works while you drive home. This is what that automation actually is, and why it changes the economics of formative assessment.
The problem isn't reading, it's timing
Be precise about why exit tickets fail, because it explains why this fix works. The value of an exit ticket decays fast. Read tonight, it reshapes tomorrow — you reteach the misconception, regroup the students who are lost, adjust the warm-up. Read Thursday, it's an autopsy. The insight is the same; the window to act on it has closed.
Manual review loses the timing race almost every time, not because teachers are lazy but because the review competes with everything else in the same exhausted hour. A scheduled task doesn't compete. It runs at 4pm whether or not you have the energy, does the synthesis, and has tomorrow's adjustment ready before you've hit the highway. The task didn't make you faster. It moved the work off your plate entirely and onto a clock that never gets tired.
The scheduled task doesn't help you read exit tickets sooner — it reads them for you, every school day at 4pm, and adapts tomorrow's plan while you drive home. The point of a handoff is that you're not in the loop anymore.
Anatomy of the scheduled task
A scheduled task is just a recurring instruction with a trigger, an action, and an output. Spelled out, the 4pm exit-ticket review looks like this:
| Part | What it is |
|---|---|
| Trigger | Every school day at 4pm — no click, no reminder |
| Input | The day's exit tickets, plus tomorrow's draft plan and the standard you're on |
| Action | Read every ticket, cluster the responses, diagnose misconceptions, map gaps to the progression |
| Adaptation | Revise tomorrow's plan against what today revealed — reteach, regroup, adjust the warm-up |
| Output | A short summary and an updated plan, waiting for you the next morning |
What makes the "action" and "adaptation" rows more than keyword-matching is the grounding underneath. Claude for Teachers connects to the Learning Commons Knowledge Graph, which carries the standards for all 50 states and the learning progressions beneath each one. So when the review finds that a cluster of students stumbled, it can place the stumble on the progression — which prerequisite competency is shaky — and adapt tomorrow accordingly. The adjustment isn't "spend more time on this"; it's "these students need the specific earlier skill shored up first." That's an instructional judgment, made on schedule, on real evidence.
Set it up once, then trust it
The work is in the setup, and the setup is a conversation. You tell Claude what you want reviewed, on what cadence, against what plan. You describe your class the way you would to a co-teacher — the standard you're on, the students you're watching, what a red flag looks like. You decide what the output should be: a two-line summary for a quick glance, or a fully revised plan you can teach as-is. Once that's defined, the task recurs on its own.
The first week, don't trust it blindly — no more than you'd trust a new student teacher's read of the room on day one. Check its analysis against your own sense of how class went. Where it's off, correct it, and the corrections carry into how it reviews the next day. Within a week or two you'll have calibrated it to your class: your standards, your students, your bar for what counts as "most of the class didn't get it." At that point the morning summary becomes something you act on with the same confidence you'd give your own late-night read — except you got your evening back to earn it.
A word on scope. Automating the review and adaptation is not the same as automating your teaching. You still decide whether to accept tomorrow's adjusted plan; you still know which students the data can't fully explain. The handoff covers the mechanical, time-sensitive synthesis — the part that was failing not for lack of skill but for lack of hours. Your judgment stays exactly where it belongs, just no longer buried under a stack of tickets at 9pm.
Why this is a handoff, not a shortcut
It's worth naming what's actually new here, because "AI reviews your exit tickets" undersells it. This isn't a faster tool you operate. It's work that carries forward without you in the loop — the defining capability of Cowork and Claude Code inside Claude for Teachers. A chatbot answers when you ask. An agent that runs on a schedule does the job while you're doing something else, which in this case is driving home. That shift, from tool you drive to work you delegate, is the through-line of the whole product, and the exit-ticket review is the cleanest example of it.
Once you've felt it work for one task, you'll see other candidates. Any recurring, evidence-in-plan-out loop is a fit: a weekly review of the running class-data folder, a Friday synthesis of the week's quizzes, a start-of-unit check of who's missing prerequisites. The class-data workflow is the manual version of the same instinct; the scheduled task is that instinct set to a clock. If you're designing the assessments that feed a loop like this, the marketplace Exam Blueprint skill is a useful companion for building the checks worth reviewing. And the broader question of what to hand off versus what to keep in your own hands is the subject of Cowork for Teachers, the companion to this piece.
The evening you get back
Strip away the mechanics and here's the trade. Formative assessment only works if you close the loop the same day. Closing that loop manually costs you an evening you rarely have to spare, which is why the loop usually stays open. A scheduled task closes it for you, on time, every school day, and leaves the results where you'll see them first thing. You didn't lower your standards for using student data well. You just stopped paying for those standards in nights.
That's the bet the Claude for Teachers series keeps making: the practices that help students are rarely the ones teachers don't know — they're the ones the week doesn't have room for. Hand the recurring, time-sensitive work to something that doesn't need an evening, and the practice finally fits. Verification is free through June 30, 2027, and if there's one task to hand off first, it's the stack of exit tickets you meant to read last night.
Part of the Claude for Teachers series. Related: Cowork for Teachers: Handing Off Work That Carries Forward · Hand Claude a Folder: Analyzing Class Data to Plan Instruction. Browse AI tutoring skills or more builder insights.